Lenovo ThinkSystem SR860 V2 Server

The Lenovo ThinkSystem SR860 V2 is a 4-socket server that features a 4U rack design with support for high-performance GPUs. The server offers technology advances, including third-generation Intel Xeon Scalable processors with Intel Optane Persistent Memory 200 Series, and scale-up capacity with up to 12TB of system memory, up to 14x PCIe slots, and up to 48x 2.5-inch drive bays.

Suggested uses: Mission critical workloads such as SAP HANA in-memory computing, transactional databases, deep learning, analytics, big data, and virtual machine density.

Figure 1. Lenovo ThinkSystem SR860 V2

The Lenovo ThinkSystem SR860 V2 provides the advanced capabilities of four of the new third-generation Intel Xeon Scalable processors plus support for four double-wide GPUs such as the NVIDIA Tesla V100S GPU. This combination gives you significant processing power in one server.

The SR860 V2 has space for 48x 2.5-inch drive bays, 24 of which can be configured as AnyBay drives - supporting SAS, SATA or NVMe drives. NVMe drives are high-speed, low-latency storage, ideal for storage tiering.

The flexible ThinkSystem SR860 V2 server supports third-generation Intel Xeon Scalable Gold or Platinum processors and can scale from two to four processors. Built for standard workloads like general business applications and server consolidation, it can also accommodate high-growth areas such as databases and virtualization. The ThinkSystem SR860 V2’s agile design permits rapid upgrades for processors and memory, and its large, flexible storage capacity helps to keep pace with data growth.

With the capability to support up to 48 DIMMs, four sockets, up to 48 drives for internal storage, support for up to eight high-performance NVIDIA Tesla T4 GPUs or four NVIDIA V100S GPUs, and a dedicated OCP 3.0 slot for 1, 10 or 25 GbE networking, the SR860 V2 provides unmatched features and capabilities in a 4U rack-mount design.

Scalability and performance

The SR860 V2 offers numerous features to boost performance, improve scalability and reduce costs: 

• Supports two or four 3rd Gen Intel Xeon Processor Scalable processors, allowing you to start with two processors and then upgrade to four when you need it.
• Supports Gold and Platinum level H and HL processors in the Intel Xeon Processor Scalable Family. Processors supported have up to 28 cores, core speeds of up to 3.9 GHz, and TDP ratings of up to 250W.
• The SR860 V2's processor expansion tray design enables you to start with two processors and 24 DIMM sockets and upgrade to four processors and 48 DIMM sockets when you need it, helping lower initial costs and manage growth.
• Up to four processors, 112 cores, and 224 threads maximize the concurrent execution of multithreaded applications.
• Enhanced inter-processor communications with six UPI connections between adjacent processors ensures increased CPU I/O throughput.
• Supports the new bfloat16 instruction set for Enhanced Deep Learning Boost
• Intelligent and adaptive system performance with Intel Turbo Boost Technology 2.0 allows processor cores to run at maximum speeds during peak workloads by temporarily going beyond processor TDP.
• Intel Hyper-Threading Technology boosts performance for multithreaded applications by enabling simultaneous multithreading within each processor core, up to two threads per core.
• Intel Virtualization Technology integrates hardware-level virtualization hooks that allow operating system vendors to better use the hardware for virtualization workloads.
• Intel Speed Select Technology supported on some processor models, enables increased core Turbo Boost frequency on specific individual cores to maximize application performance.
• Enhanced Intel Deep Learning Boost provides built-in acceleration with the Vector Neural Network Instruction set (VNNI) and new bloat16 instructions. These are designed to deliver significant, more efficient Deep Learning (Inference) acceleration for high-performance Artificial Intelligence (AI) workloads.
• Intel Advanced Vector Extensions 512 (AVX-512) enable acceleration of enterprise-class workloads, including databases and enterprise resource planning (ERP).
• Support for up to 48 TruDDR4 memory DIMMs operating at up to 3200 MHz means you have the fastest available memory subsystem and memory capacity of up to 12 TB with 48x 256 GB 3DS RDIMMs.
• Supports configurations of 2 DIMMs per channel to operate at the 3200 MHz rated speed of the memory DIMMs.
• Supports the new Intel Optane Persistent Memory 200 Series for advanced in-memory database applications, dense-virtualization; up to 24 PMem Modules can be installed in conjunction with regular system memory. PMem modules are up to 512 GB each, for a total of up to 12 TB of Persistent Memory (3 TB per processor).
• The use of solid-state drives (SSDs) instead of, or along with, traditional spinning drives (HDDs), can improve I/O performance. An SSD can support up to 100 times more I/O operations per second (IOPS) than a typical HDD.
• Up to 48x 2.5-inch drive bays -- supporting combinations of SAS or SATA HDDs, SAS or SATA SSDs, and NVMe PCIe SSDs -- provide a flexible and scalable all-in-one platform to meet your increasing demands. Up to 24x NVMe drives are supported, maximizing drive I/O performance in terms of throughput, bandwidth, and latency.
• The server has a dedicated industry-standard OCP 3.0 small form factor (SFF) slot, with a PCIe 3.0 x16 interface, supporting a variety of Ethernet network adapters. Simple-swap mechanism with thumbscrews and pull-tab enables tool-less installation and removal of the adapter. Supports shared BMC network sideband connectivity to enable out-of-band systems management.
• Up to 14 PCIe 3.0 slots in addition to the OCP 3.0 Ethernet slot to maximize I/O capabilities.
• New high-speed RAID controllers from Broadcom provide 12 Gb SAS connectivity to the drive backplanes. A variety of RAID adapters are available, with cache up to 8 GB and support for 32 drives on a single controller. The RAID 940-16i 8GB Flash adapter supports RAID volumes with NVMe, SAS or SATA drives.
• Support for four high-performance double-wide GPUs such as the NVIDIA Tesla V100S, or eight single-wide GPUs such as the NVIDIA T4 GPU. GPUs adds additional processing power to the server.
• Supports up to two externally accessible 7mm hot-swap drives with RAID functionality, in addition to the 48 front drive bays. These 7mm drives are ideal for operating system boot functions.
• As an alternative to the 7mm drives, the server supports the Lenovo patented-design M.2 adapter for convenient operating system boot functions. Available M.2 adapters support either one M.2 drive or two M.2 drives in a RAID 1 configuration for boot drive performance and reliability.
• Integrated SATA controller with Software RAID (Intel RSTe) support for up to 8x drives with basic RAID functionality with no additional adapter needed.
• Supports Intel VROC (Virtual RAID on CPU) which enables basic RAID functionality on the onboard NVMe ports of the server, with no additional adapter needed. This feature enables RAID on NVMe drives without the need for a separate RAID adapter.

Availability and serviceability

The SR860 V2 provides many features to simplify serviceability and increase system uptime:

• The server offers Single Device Data Correction (SDDC, also known as Chipkill), Adaptive Double-Device Data Correction (ADDDC, also known as Redundant Bit Steering or RBS) and memory mirroring for redundancy in the event of a non-correctable memory failure.
• The server offers hot-swap drives, supporting RAID redundancy for data protection and greater system uptime.
• Available M.2 RAID Boot Adapter supports RAID-1 which can enable two SATA M.2 drives to be configured as a redundant pair.
• Available pair of 7mm hot-swap drives support RAID-1 which can enable two SATA or two NVMe drive to be configured as a redundant pair.
• The server has up to four hot-swap redundant power supplies and eight N+1 redundant fans to provide availability for business-critical applications.
• The power-source-independent light path diagnostics uses LEDs to lead the technician to failed (or failing) components, which simplifies servicing, speeds up problem resolution, and helps improve system availability.
• An integrated diagnostics panel with LCD display (optional in China, standard in other markets) provides more detailed diagnostics by displaying all error messages and VPD data needed for a service call, thereby aiding with problem resolution and system uptime.
• Proactive Platform Alerts (including PFA and SMART alerts): Processors, voltage regulators, memory, internal storage (SAS/SATA HDDs and SSDs, NVMe SSDs, M.2 storage, flash storage adapters), fans, power supplies, RAID controllers, server ambient and subcomponent temperatures. Alerts can be surfaced through the XClarity Controller to managers such as Lenovo XClarity Administrator, VMware vCenter, and Microsoft System Center. These proactive alerts let you take appropriate actions in advance of possible failure, thereby increasing server uptime and application availability.
• Solid-state drives (SSDs) offer more reliability than traditional mechanical HDDs for greater uptime.
• The built-in XClarity Controller continuously monitors system parameters, triggers alerts, and performs recovery actions in case of failures, to minimize downtime.
• Built-in diagnostics in UEFI, using Lenovo XClarity Provisioning Manager, speed up troubleshooting tasks to reduce service time.
• Lenovo XClarity Provisioning Manager collects and saves service data to USB key drive or remote CIFS share folder, for troubleshooting and to reduce service time.
• Auto restart in the event of a momentary loss of AC power (based on the power policy setting in the XClarity Controller service processor)
• Support for the XClarity Administrator Mobile app running on a supported smartphone or tablet and connected to the server through the front USB 2.0 port, enables additional local systems management functions.
• Three-year or one-year customer-replaceable unit and onsite limited warranty, 9 x 5 next business day. Optional service upgrades are available.

The following figure shows the architectural block diagram of the SR860 V2, showing the major components and their connections.

Figure 7. SR860 V2 system architectural block diagram

The following table lists the standard specifications.

The SR860 V2 servers are shipped with the following items:

• Documentation flyer
• Rail kit (model dependent)
• Power cords (model and region dependent)

ThinkSystem SR860 V2 models can be configured by using the Lenovo Data Center Solution Configurator (DCSC).

Configure-to-order (CTO) models are used to create models with factory-integrated server customizations. For CTO models, two types of base CTO models are available for the SR860 V2 as listed in the columns in the following table:

• General purpose base CTO models are for general business (non-HPC) and is selectable by choosing General Purpose mode in DCSC.
• AI and HPC base models are intended for Artificial Intelligence (AI) and High Performance Computing (HPC) configurations and solutions are enabled using the AI & HPC Hardware - ThinkSystem Hardware mode in DCSC. These configurations, along with Lenovo EveryScale Solutions, can also be built using System x and Cluster Solutions Configurator (x-config). Tip: Some HPC and AI models are not listed in DCSC and can only be configured in x-config.

Preconfigured server models may also be available for the SR860 V2, however these are region-specific; that is, each region may define their own server models, and not all server models are available in every region.

The following table lists the base CTO models of the ThinkSystem SR860 V2 server.

The following tables list the available models, grouped by region:

• Models for AP
• Models for EMEA

The SR860 V2 supports Gold and Platinum level processors in the third-generation Intel Xeon Scalable Processor family.

Topics in this section:

• Memory capacity of processors
• Processor features
• Two-processor configurations
• UEFI operating modes

The server supports two or four processors.

• Processors 1 and 2 are installed in sockets on the system board.
• Processors 3 and 4 are installed on the Processor and Memory Expansion Tray.

The Processor and Memory Expansion Tray has a sliding and rolling lift mechanism that requires only a one-handed operation to insert or remove from the server. It has a toolless design for superior usability, serviceability and reliability. 

Figure 8. SR860 V2 Processor and Memory Expansion Tray

The ordering information for the Processor and Memory Expansion Tray is shown in the following table.

The following processor quantities are supported:

• Two processors, installed in sockets 1 and 2, without the Processor and Memory Expansion Tray
• Four processors, installed in all four sockets

Note: Configurations of one or three processors are not supported.

The table below lists the supported third-generation Intel Xeon Scalable processors.

All supported processors have the following characteristics:

• Third-generation Intel Xeon Scalable processors (formerly codenamed "Cooper Lake" or CPX6)
• 14 nm process technology
• Six DDR4 memory channels
• 48 PCIe 3.0 I/O lanes
• 1 MB L2 cache
• 1.375 MB or more L3 cache per core
• Six Intel Ultra Path Interconnect (UPI) links at 10.4 GT/s
• Intel Hyper-Threading Technology
• Intel Turbo Boost Technology 2.0
• Intel Advanced Vector Extensions 512 (AVX-512)
• Vector Neural Network Instructions (VNNI)
• Enhanced Deep Learning Boost (bfloat16)
• Intel Speed Select support (SST-TF and SST-CP)
• Intel Persistent Memory 200 Series support

The processors include a suffix letter in the processor model number:

• H: 4-Socket capable processor, supports memory up to 1.125TB per processor
• HL: 4-Socket capable, with large memory tier (supports memory up to 4.5TB per processor)

The four processors are connected together using a mesh topology. A mesh topology allows all four processors to be connected together which improves the performance of processor-to-processor communications. The SR860 V2 implements a mesh topology using 6 UPI links. The following figure shows how the SR860 V2 mesh topology is different to the ring topology of the SR850 server and the mesh topology of the SR850P server.

Figure 9. Ring toplogy (SR850) versus 3-UPI Mesh topology (SR850P) versus 6-UPI Mesh topology (SR860 V2)

Memory capacity of processors

Third-generation Xeon Scalable processors are limited to the amount of memory they can address, as follows:

• Processors with an L suffix (eg 8360HL): Up to 4.5 TB per processor (4608 GB per processor)
• Processors without an L suffix (eg 8360H): Up to 1.125 TB per processor (1152 GB per processor)

The calculation of the total memory per processor includes both the system memory DIMMs and the persistent memory modules installed in the server.

For example, these configurations do not require an L processor:

• A configuration using 12x 64GB DIMMs per processor, which is a total of 768GB per processor
• A configuration using 6x 64GB DIMMs + 6x 128GB PMem per processor, which is a total of 1.125TB per processor

For example, these configurations must have an L processor installed:

• A configuration using 12x 128GB DIMMs per processor, which is a total of 1.5TB per processor
• A configuration using 6x 32GB DIMMs + 6x 256GB PMem, which is a total of 1.6875TB per processor

Processor features

The following table lists the features of the supported third-generation Intel Xeon processors.

* L3 cache is 1.375 MB per core for all processors except the Platinum 8356H which has 4.47 MB per core

Two-processor configurations

The SR860 V2 can be used with only two processors installed. Most core functions of the server (including the XClarity Controller) are connected to processors 1 and 2 as shown in the System architecture section.

With only two processors, the server has the following capabilities:

• 24 memory DIMMs for a 3TB maximum
• Up to 12 PMem modules
• Slots 1-4 and 9-12 are not available
• GPUs not supported
• NVMe Switch Adapters not supported
• Up to 8x NVMe drives
  • If the 3-slot riser is used for slots 5-7 (Riser 1), then NVMe drives are not supported

Some drive bay configurations require 4 processors. See the table in the Supported drive bay combinations section for details.

The following table lists the feature codes that allow you to specify the mode you wish to preset in the factory for CTO orders.

The preset modes for the SR860 V2 are as follows:

• Maximum Performance Mode (feature BFYB): Achieves maximum performance but with higher power consumption and lower energy efficiency.
• Minimal Power Mode (feature BFYC): Minimize the absolute power consumption of the system.
• Efficiency Favoring Power Savings Mode (feature BFYD): Maximize the performance/watt efficiency with a bias towards power savings. This is the favored mode for SPECpower benchmark testing, for example.
• Efficiency Favoring Performance Mode (feature BFYE): Maximize the performance/watt efficiency with a bias towards performance. This is the favored mode for Energy Star certification, for example.

For details about these preset modes, and all other performance and power efficiency UEFI settings offered in the SR860 V2, see the paper "Tuning UEFI Settings for Performance and Energy Efficiency on Intel Xeon Scalable Processor-Based ThinkSystem Servers", available from https://lenovopress.lenovo.com/lp1477.

The SR860 V2 uses Lenovo TruDDR4 memory and supports 12 DIMMs per processor, which corresponds to 48 DIMMs with four processors installed. Each processor has six memory channels with two DIMMs per channel. With 256 GB 3DS RDIMMs installed, a 4-socket server supports a total of 12 TB of system memory.

The SR860 V2 also supports Intel Optane Persistent Memory 200 Series, as described in the Persistent Memory section.

Memory operates at up to 3200 MHz at two DIMMs per channel, depending on the processor model selected. If the processor selected has a memory bus speed of 2666 MHz or 2933 MHz, then all DIMMs will operate at that speed.

The following table lists the memory options that are available for the server.

Lenovo TruDDR4 memory uses the highest quality components that are sourced from Tier 1 DRAM suppliers and only memory that meets the strict requirements of Lenovo is selected. It is compatibility tested and tuned to maximize performance and reliability. From a service and support standpoint, Lenovo TruDDR4 memory automatically assumes the system warranty, and Lenovo provides service and support worldwide.

The following rules apply when selecting the memory configuration:

• Only specific DIMM quantities are supported:
  • For 16GB and 256GB DIMMs, only quantities of 1, 4, 6 or 12 DIMMs per processor are supported
  • For 32GB, 64GB and 128GB DIMMs, only quantities of 1, 4, 6, 8 or 12 DIMMs per processor are supported
• The server supports RDIMMs and 3DS RDIMMs; UDIMMs and LRDIMMs are not supported
• Mixing RDIMMs and 3DS RDIMMs is not supported
• Mixing 128GB and 256GB 3DS RDIMMs is not supported
• Mixing x4 and x8 DIMMs is supported
• Additional restrictions may apply when mixing DIMM capacities – use the Memory Wizard to determine if your memory configuration is supported.
• If 128GB or 256GB 3DS DIMMs are selected, at most 24 front drive bays can be configured
• If 256GB 3DS DIMMs are selected, at most 16 front drive bays can be configured

For best performance, consider the following:

• Populate memory DIMMs in quantities of 6 or 12 per processor, so that all memory channels are used.
• Populate memory channels so they all have the same total memory capacity.
• Ensure all memory controllers on a processor socket have the same DIMM configuration.
• All processor sockets on the same physical server should have the same DIMM configuration.

See the Lenovo Press article "RAS Features of the Lenovo ThinkSystem Intel Servers" for more information about memory RAS features: https://lenovopress.lenovo.com/lp1711-ras-features-of-the-lenovo-thinksystem-intel-servers

If memory channel mirroring is used, then DIMMs must be installed in pairs or sets of three (minimum of one pair or set of three per processor), and all DIMMs in the pair or set of three must be identical in type and size. 50% of the installed capacity is available to the operating system.

The SR860 V2 server supports Intel Optane Persistent Memory 200 Series, a new class of memory and storage technology explicitly architected for data center usage. Persistent memory is an innovative technology that delivers a unique combination of affordable large memory capacity and persistence (non-volatility). It offers significantly lower latency than fetching data from SSDs, even NVMe SSDs, and offers higher capacities than system memory.

Persistent memory technology can help boost the performance of data-intensive applications such as SAP HANA. When data is stored closer to the processor on nonvolatile media, applications can see significant overall improvement in performance.

The following table lists the ordering information for the supported persistent memory modules.

The following are the requirements when installing persistent memory (PMem) modules:

• Only App Direct Mode is supported. Memory Mode and Mixed Mode are not supported.
• All processors that the SR860 V2 supports also support the PMem modules listed in the table below.
• All PMem modules operate at 2666 MHz. Installed memory DIMMs will also operate at 2666 MHz.
• All installed PMem modules must be the same size. Mixing PMem modules of different capacities is not supported.
• Maximum 6 PMem modules per processor (install 1 in each memory channel).
• For each memory channel with both a PMem module and a memory DIMM installed, the PMem module is installed in channel slot 1 (DIMM1, closer to the processor) and the DIMM is installed in channel slot 0 (DIMM0).
• Both interleaved and non-interleaved modes are supported.
• To maximize performance, balance all memory channels
• Memory mirroring is not supported with PMem modules installed
• At most 24x 2.5-inch drives can be installed when any PMem modules are installed. The lower backplanes (BP4, BP5 and BP6) must remain unpopulated.
• Ambient temperature cannot exceed 30°C
• Only the following operating systems support PMem modules in the SR860 V2:
  • SUSE Linux Enterprise Server 15 SP2 or later
  • SUSE Linux Enterprise Server 15 Xen SP2 or later

In App Direct Mode, only the combinations of RDIMMs and PMem modules listed in the following table are supported. A required RDIMM:PMem ratio of between 1:1 and 1:4 must be met.

PMem modules offer the following memory protection technologies:

• ECC
• Device failure recovery - SDDC
• Die sparing - DDDC
• Patrol scrubbing

For more information, see the Intel Optane Persistent Memory 200 Series product guide, https://lenovopress.com/LP1380

The SR860 V2 supports up to 48x 2.5-inch SAS/SATA drive bays, up to 24 of which can be NVMe or AnyBay drive bays instead. All 48x drive bays are hot-swap and all front-accessible. The server also supports internal M.2 drives (one or two, installed in an adapter), or rear-accessible hot-swap 7mm SSDs.

In this section:

• NVMe drive support
• Front drive bays
• Supported drive bay combinations
• Controller selections
• Field upgrades
• Cable kits for field upgrades
• M.2 drives
• 7mm drives
• SED encryption key management with ISKLM

NVMe drive support

The SR860 V2 supports up to 24x NVMe drives to maximize storage performance.

• Up to 16 NVMe drives with direct connections to the processors
  • No oversubscription (each x4 drive has a full x4 (4 lanes) connection to the processor)
  • 4 drives are connected to each processor for balanced performance
  • 16 NVMe drives precludes the use of the 3-slot riser (Slots 5-7). See I/O expansion options for details.
• Up to 24 NVMe drives using direct connections plus NVMe switch adapters or retimers
  • No oversubscription (each x4 drive has a full x4 (4 lanes) connection to the processor)
  • Each NVMe Retimer adapter or 1610-8P switch adapter has 4 drives attached (no oversubscription)

The specifics of these configurations are covered in the Supported drive bay combinations and Controller selections sections.

The tables in those sections indicate the number of NVMe drives in each configuration plus the subscription ratio. The subscription ratio is the number of PCIe lanes from the processor compared to the number of lanes to the drives. A ratio of 1:1 means all drives get the full number of lanes they need to maximize drive performance (currently 4 lanes per drive). A ratio of 1:2 means each drive only gets the half the bandwidth from the processor.

In addition, the server supports two 7mm NVMe drives for use as boot drives. These two drives are connected via separate RAID controller connected to a single PCIe 3.0 x2 host interface. See the 7mm drives section for details.

Tri-Mode support

Ordering information for the backplanes is listed in the following table.

Supported drive bay combinations

The following table shows the supported drive bay combinations - SAS/SATA, AnyBay or NVMe drives. The table lists the backplanes required for each drive bay combination.

Some configurations require 4 processors be installed. This is noted in the table.

The table also indicates whether the combination permits the use of the 3-slot riser in Riser 1 (slots 5-7). Those configurations that don't support the 3-slot riser have 16 or more AnyBay or NVMe drives, or are an upgrade path for 16 drives. Such configurations can use the 2-slot riser (slots 5 & 6) instead.

The choice of storage controller for each configuration is listed in the Controller selections section.

§ The text in parenthesis refers to the subscription ratio. See the NVMe support section for details.
‡ Configurations that have 16 AnyBay or NVMe drives (or are an upgrade path for 16 drives) do not support the use of the 3-Slot Riser (Riser Z, slots 5-7)
* Depending on the controllers used in this configuration, the 3-Slot Riser may not be supported. See Controller selections section for details.

Controller selections

This section helps you determine with storage adapter are supported for your desired drive bay configuration.

In the tables, the controllers are grouped as follows:

• RAID/HBA 8i corresponds to any of the following:
  • ThinkSystem RAID 530-8i PCIe 12Gb Adapter, 7Y37A01082
  • ThinkSystem RAID 5350-8i PCIe 12Gb Adapter, 4Y37A72482
  • ThinkSystem RAID 540-8i PCIe Gen4 12Gb Adapter, 4Y37A78834
  • ThinkSystem RAID 930-8i 2GB Flash PCIe 12Gb Adapter, 7Y37A01084
  • ThinkSystem RAID 9350-8i 2GB Flash PCIe 12Gb Adapter, 4Y37A72483
  • ThinkSystem RAID 940-8i 4GB Flash PCIe Gen4 12Gb Adapter, 4Y37A09728
  • ThinkSystem RAID 940-8i 8GB Flash PCIe Gen4 12Gb Adapter, 4Y37A09729
  • ThinkSystem 430-8i SAS/SATA 12Gb HBA, 7Y37A01088
  • ThinkSystem 4350-8i SAS/SATA 12Gb HBA, 4Y37A72480
  • ThinkSystem 440-8i SAS/SATA PCIe Gen4 12Gb HBA, 4Y37A78601
• RAID/HBA 16i corresponds to any of the following:
  • ThinkSystem RAID 530-16i PCIe 12Gb Adapter, 4Y37A09727
  • ThinkSystem RAID 930-16i 4GB Flash PCIe 12Gb Adapter, 7Y37A01085
  • ThinkSystem RAID 9350-16i 4GB Flash PCIe 12Gb Adapter, 4Y37A72485
  • ThinkSystem RAID 940-16i 4GB Flash PCIe Gen4 12Gb Adapter, 4Y37A78600
  • ThinkSystem RAID 940-16i 8GB Flash PCIe Gen4 12Gb Adapter, 4Y37A09730
  • ThinkSystem 430-16i SAS/SATA 12Gb HBA, 7Y37A01089
  • ThinkSystem 4350-16i SAS/SATA 12Gb HBA, 4Y37A72481
  • ThinkSystem 440-16i SAS/SATA PCIe Gen4 12Gb HBA, 4Y37A78602
• RAID 32i corresponds to the following:
  • ThinkSystem RAID 940-32i 8GB Flash PCIe Gen4 12Gb Adapter, 4Y37A09733
• Retimer/1610-8P corresponds to either of the following:
  • ThinkSystem 1610-8P NVMe Switch Adapter, 4Y37A09764 (only 4 ports are used)
  • ThinkSystem 4-Port PCIe Gen4 NVMe Retimer Adapter, 4C57A65446
• OB SATA (onboard SATA) corresponds to the following in CTO orders:
  • On Board SATA Software RAID Mode, feature AVV0
• OB NVMe (onboard NVMe) corresponds to the following in CTO orders:
  • Non RAID NVMe, feature BC4V
  • Intel VROC (VMD NVMe RAID) Intel SSD Only, feature B9X7
  • Intel VROC (VMD NVMe RAID) Premium, feature B96G

All configurations also support the 7mm rear drive bays (which can can be either SATA or NVMe), or they can support M.2 drives. See the 7mm drives and M.2 drives sections for details. 7mm drive support and M.2 drive support is mutually exclusive.

The following table lists the supported drive bay combinations, plus the list of supported controller combinations supported by each of those drive bay combinations. Information about the controllers can be found in the Controllers for internal storage section.

Some configurations require 4 processors be installed. This is noted in the table.

The table also indicates whether the combination permits the use of the 3-slot riser (slots 5-7). Those configurations that don't support the 3-slot riser have 16 or more AnyBay or NVMe drives, or are an upgrade path for 16 drives. Such configurations can use the 2-slot riser (slots 5 & 6) instead.

Field upgrades

The SR860 V2 is orderable without drive bays, allowing you to add a backplane, cabling and controllers as field upgrades. The server also supports upgrading some configurations by adding additional front drive bays, adapters, and cables.

Each row of the following table represents a supported upgrade path. Any other upgrades may not be supported.

To add drive bays you will need to order both drive backplanes and cable kits. Backplane kits do not include cables.

* Abbreviations such as "RAID/HBA 16i Adapter" refer to a group of supported adapters; these are described in the Controller selections section

Supported drives are listed in the Internal drive options section.

For field upgrades, the SR860 V2 also requires an additional M.2 cable kit. Ordering information is listed in the following table.

For further details about M.2 components, see the ThinkSystem M.2 Drives and M.2 Adapters product guide:
https://lenovopress.com/lp0769-thinksystem-m2-drives-adapters

Ordering information for the 7mm drive bays is in the following table.

Each drive enclosure includes an integrated controller providing RAID functions.

The IBM Security Key Lifecycle Manager software is available from Lenovo using the ordering information listed in the following table.

The SR860 V2 supports offers a variety of controller options for internal drives:

• For 2.5-inch drives:
  • Onboard SATA ports with software RAID support (Intel VROC SATA RAID, formerly known as Intel RSTe)
  • Onboard NVMe ports with software RAID support (Intel VROC NVMe RAID)
  • RAID adapters and HBAs for SAS/SATA drives (RAID 940-16i also supports NVMe)
  • NVMe Switch adapter for NVMe drives
• For 7mm drive bays in the rear of the server (see the 7mm drives section)
  • SATA controller integrated into the 7mm drive bay enclosure
  • NVMe controller integrated into the 7mm drive bay enclosure
• For M.2 drives internal to the server (see M.2 drives section)
  • SATA controller integrated on the M.2 SATA 2-Bay RAID Enablement Kit
  • NVMe controller integrated on the M.2 NVMe 2-Bay RAID Enablement Kit

The following table lists the controllers and adapters used for the internal 2.5-inch drive bays of the SR860 V2 server.

• E810 Ethernet and X350 RAID/HBAs: The use of both an Intel E810 network adapter and an X350 HBA/RAID adapter (9350, 5350 and 4350) is currently not supported in ThinkSystem servers. For details see Support Tip HT513226. Planned support for this combination of adapters is 2Q/2024.

VROC SATA RAID (RSTe) is available and supported with all SATA drives. It offers a 6 Gb/s connection to each drive and on the SR860 V2 implements RAID levels 0, 1, 5, and 10. RAID 1 is limited to 2 drives per array, and RAID 10 is limited to 4 drives per array. Hot-spare functionality is also supported.

VROC NVMe RAID offers RAID support for any NVMe drives directly connected to the ports on the server's system board or via adapters such as NVMe retimers or NVMe switch adapters. On the SR860 V2, RAID levels implemented are based on the VROC feature selected as indicated in the following table. RAID 1 is limited to 2 drives per array, and RAID 10 is limited to 4 drives per array. Hot-spare functionality is also supported.

The SR860 V2 supports the VROC NVMe RAID offerings listed in the following table. The VROC Intel SSD Only offering only supports RAID on Intel branded NVMe SSDs; non-Intel branded NVMe SSDs cannot be configured in a RAID array.

Configuration notes:

• If a feature code is ordered in a CTO build, the VROC functionality is enabled in the factory. For field upgrades, order a part number and it will be fulfilled as a Feature on Demand (FoD) license which can then be activated via the XCC management processor user interface.

For specifications about the RAID adapters and HBAs supported by the SR860 V2, see the ThinkSystem RAID Adapter and HBA Comparison, available from:
https://lenovopress.com/lp1288-lenovo-thinksystem-raid-adapter-and-hba-reference#sr860-v2-support=SR860%2520V2

For details about these adapters, see the relevant product guide:

• SAS HBAs: https://lenovopress.com/servers/options/hba
• RAID adapters: https://lenovopress.com/servers/options/raid

The following tables list the drive options for internal storage of the server.

The server does not support any internal backup units, such as tape drives or RDX drives. 

The server does not support an internal optical drive.

An external USB optical drive is available, listed in the following table.

The SR860 V2 has up to 14 PCIe 3.0 slots plus an OCP 3.0 slot reserved for an Ethernet adapter.

The following figure shows the locations of the PCIe slots.

Figure 12. SR860 V2 slot numbering

The slots are grouped on risers or directly on the system planar, as shown in the figure:

• Riser 2 (x32 riser connection to Expansion Tray):
  • Slot 1: PCIe x8 FHFL (physically x16) (not present if slot 2 is PCIe x16) (connects to CPU 4)
  • Slot 2: PCIe x8 FHFL (physically x16) or PCIe x16 FHFL double-wide (CPU 4)
  • Slot 3: PCIe x8 FHFL (physically x16) (not present if slot 4 is PCIe x16) (CPU 4)
  • Slot 4: PCIe x8 FHFL (physically x16) or PCIe x16 FHFL double-wide (CPU 4)
• Riser 1 (x16 riser slot on system planar with two additional x16 cabled connections):
  • Slot 5: PCIe x16 FHHL (CPU 2) or PCIe x8 FHHL with 2-slot riser (CPU 2)
  • Slot 6: PCIe x16 FHHL (CPU 1) or PCIe x8 FHHL with 2-slot riser (CPU 1)
  • Slot 7: PCIe x16 FHHL (not present if 2-slot riser is selected) (CPU 2)*
• Slot 8: OCP 3.0 slot with PCIe x16 connection (CPU 2)
• Riser 3 (x32 riser connection to Expansion Tray):
  • Slot 9: PCIe x8 FHFL (physically x16) (not present if slot 10 is PCIe x16) (CPU 3)
  • Slot 10: PCIe x8 FHFL (physically x16) or PCIe x16 FHFL double-wide (CPU 3)
  • Slot 11: PCIe x8 FHFL (physically x16) (not present if slot 12 is PCIe x16) (CPU 3)
  • Slot 12: PCIe x8 FHFL (physically x16) or PCIe x16 FHFL double-wide (CPU 3)
• System planar slots:
  • Slot 13: PCIe x16 LP (CPU 1)
  • Slot 14: PCIe x8 LP (physically x8) (CPU 1)
  • Slot 15: PCIe x8 LP (physically x8) (CPU 1)

Riser 2 and Riser 3 support two different riser card selections:

• 4x PCIe 3.0 x8 slots (physically x16) supporting single-wide GPUs and other FHFL adapters - Feature code BAW9
• 2x PCIe 3.0 x16 slots supporting double-wide GPUs and other FHFL adapters - Feature code BAW8

Riser 1 supports two different riser card selections:

• 3x PCIe 3.0 x16 slots, slots 5, 6, & 7, supporting FHHL adapters - Feature code BAW7. The connections of these slots is shown in the System architecture section:
  • Slot 5 connects to CPU 2 via the riser slot
  • Slot 6 connects to CPU 1 via a cable to a connector on the system board
  • Slot 7 connects to CPU 2 via a cable to a connector on the system board

  Note: The use of this riser precludes the use of 8 onboard NVMe drive connections, since they use the same connectors.

• 2x PCIe 3.0 x8 slots (physically x16), slots 5 & 6, supporting FHHL adapters - Feature code BAW6.
  • Slot 5 connects to CPU 2 via the riser slot
  • Slot 6 connects to CPU 2 via the riser slot

  Note: This riser still permits the use of the 8 onboard NVMe drive connections that the 3-slot riser consumes.

The riser cards supported are listed in the following table.

The server has a dedicated OCP 3.0 SFF slot with PCIe 3.0 x16 host interface. See Figure 3 for the location of the OCP slot.

The following table lists the supported OCP adapters. One port can optionally be shared with the XCC management processor for Wake-on-LAN and NC-SI support. Only one OCP card can be installed in the server.

The following table lists additional supported network adapters that can be installed in the regular PCIe slots.

For details about these adapters, see the relevant product guide:

• Ethernet adapters: https://lenovopress.com/servers/options/ethernet
• InfiniBand adapters: https://lenovopress.com/servers/options/infiniband

The following table lists the Fibre Channel HBAs supported by the server.

For more information, see the list of Lenovo Press Product Guides in the Host bus adapters category:
https://lenovopress.com/servers/options/hba

The following table lists SAS HBAs and RAID adapters supported by the server for use with external storage.

The RAID 930-8e, like other RAID 930 and RAID 940 adapters, uses a flash power module (supercap) and the server supports up to four supercaps. The number of 930-8e RAID adapters supported is based on how many supercaps can be installed in the server. For example, if your configuration uses two RAID 930/940 adapters for internal storage, then you can only install two RAID 930-8e adapters, since there is only space for four supercaps total. 

For details about these adapters, see the relevant product guide:

• SAS HBAs: https://lenovopress.com/servers/options/hba
• RAID adapters: https://lenovopress.com/servers/options/raid

The following table compares the specifications of the external SAS HBAs and RAID adapters.

* SAS HBAs support SEDs (self-encrypting drives) by using software on the server and simply passing SED commands through the HBA to the drives. SED support by RAID controllers is provided using the built-in MegaRAID SafeStore functionality of the adapter.

To compare these adapters, see the ThinkSystem RAID Adapter and HBA Comparison Reference:
https://lenovopress.com/lp1288-thinksystem-raid-adapter-and-hba-reference#internal-or-external-ports=External

All supported flash storage adapters are now withdrawn from marketing.

The SR860 V2 supports the graphics processing units (GPUs) listed in the following table:

For details about these GPUs, see the ThinkSystem and ThinkAgile GPU Summary:
https://lenovopress.com/lp0768-thinksystem-thinkagile-gpu-summary

The following rules apply when using GPUs:

• Select the correct riser, as listed in the table below.
• Installed GPUs must be identical

• Some NVIDIA A Series GPUs are available as two feature codes, one with a CEC chip and one without a CEC chip (ones without the CEC chip have "w/o CEC" in the name). The CEC is a secondary Hardware Root of Trust (RoT) module that provides an additional layer of security, which can be used by customers who have high regulatory requirements or high security standards. NVIDIA uses a multi-layered security model and hence the protection offered by the primary Root of Trust embedded in the GPU is expected to be sufficient for most customers. The CEC defeatured products still offer Secure Boot, Secure Firmware Update, Firmware Rollback Protection, and In-Band Firmware Update Disable. Specifically, without the CEC chip, the GPU does not support Key Revocation or Firmware Attestation. CEC and non-CEC GPUs of the same type of GPU can be mixed in field upgrades.

GPUs other than the NVIDIA T4 require an auxiliary power cable. For all other GPUs, the necessary auxiliary power cables are automatically selected as part of any factor (CTO) order. However when ordering these GPUs as field upgrades, you will need to also order the power cable separately. One part number is needed per GPU.

The server has eight 60 mm hot-swap variable-speed fans at the front of the server and all eight fans are standard in all models. The server offers N+1 redundancy, meaning that one fan can fail and the server still operates normally.

In addition, each of the two upper riser cages has two integrated dual-rotor fans. The pair of fans offer rotor redundancy - a riser cage can tolerate a single rotor failure and continue to operate.

Each power supply also includes an integrated fan.

Figure 13. SR860 V2 front system fans

When servicing the fan modules, you remove the modules from the top of the unit (hot-swap). In the case of fan modules 2 and 5, two fans are removed at the same time. Although the server supports N+1 redundancy (that is, supporting the failure of 1 fan while maintaining server operation), the server supports the removal of two fans for the time it takes to undertake a fan replacement: remove the module, replace the fan, reinsert the module.

The server supports two types of fan modules:

• Standard fan modules: Bottom fan is a single-rotor fan, top fan is a dual-rotor fan
• Performance fan modules: Both bottom and top fans are dual-rotor fans

Fan module types cannot be mixed. All six modules must be either all standard fan modules, or all performance fan modules.

Performance fan modules are required for certain configurations:

• CPU > 165W TDP
• CPU suffix is HL (large memory support)
• 128GB or 256GB 3DS RDIMMs installed
• Persistent Memory installed

The following table lists the CTO ordering information for the fan modules.

* For field upgrades, see the next table

The server supports upgrading all standard fan modules to performance fan modules, using the kit listed in the following table.

The server supports up to four redundant hot-swap power supplies. Redundancy can be configured as N+1 or N+N.

The 750W Platinum and 1100W power supplies are auto-sensing and support both 110V AC (100-127V 50/60 Hz) and 220V AC (200-240V 50/60 Hz) power. The 750 W Titanium and 1800 W power supplies only supports 220V AC power. For China customers, all power supplies support 240V DC.

Power supply options do not include a line cord. For server configurations, the inclusion of a power cord is model dependent. Configure-to-order models can be configured without a power cord if desired.

Zero-output mode: When Zero-output mode (also known as Standby mode or Cold Redundancy mode) is configured in XCC and the server power load is sufficiently low, one of the installed power supplies enters into the Standby state while the other one delivers entire load. When the power load increases, the standby power supply will switch to Active state to provide sufficient power to the server. Zero-output mode can be enabled or disabled in the XClarity Controller web interface, Server Configuration > Power Policy. If you select Disable, then both power supplies will be in the Active state.

The SR860 V2 contains an integrated service processor, XClarity Controller (XCC), which provides advanced control, monitoring, and alerting functions. The XCC is based on the Pilot4 XE401 baseboard management controller (BMC) using a dual-core ARM Cortex A9 service processor.

Topics in this section:

• Local management
• System status with XClarity Mobile
• Remote management
• Lenovo XClarity Provisioning Manager
• Lenovo XClarity Administrator
• Lenovo XClarity Essentials
• Lenovo XClarity Energy Manager
• Lenovo Capacity Planner

The SR860 V2 includes an pull-out integrated diagnostics panel with LCD display (optional for customers in China, standard for all other markets) as shown in the following figure.

Figure 15. Integrated diagnostics panel with LCD display

The Integrated diagnostics panel allows quick access to system status, firmware, network, and health information. The LCD display on the panel and the function buttons give you access to the following information:

• Active alerts
• Status Dashboard
• System VPD: machine type & mode, serial number, UUID string
• System firmware levels: UEFI and XCC firmware
• XCC network information: hostname, MAC address, IP address, DNS addresses
• Environmental data: Ambient temperature, CPU temperature, AC input voltage, estimated power consumption
• Active XCC sessions
• System reset action

The front operator panel with LCD display can be ordered as listed in the following table.

* Only available configure-to-order or with predefined models

Light path diagnostics

The server offers light path diagnostics. If an environmental condition exceeds a threshold or if a system component fails, XCC lights LEDs inside the server to help you diagnose the problem and find the failing part. The server has fault LEDs next to the following components:

• Each processor
• Each memory DIMM
• Each drive bay
• Each system fan
• Each power supply

A Light Path button is located on the system board, adjacent to Riser 1 as shown in the Internal view in the Components and connectors section. When you press this button, the LED associated with the failed component lights up. The button is attached to a supercap so that the LED lights up even if AC power is removed.

Information tab

The front of the server also houses an information pull-out tab (also known as the network access tag). See Figure 2 for the location. A label on the tab shows the network information (MAC address and other data) to remotely access XClarity Controller.

The steps to connect the mobile device are as follows:

1. Enable USB Management on the server, by holding down the ID button for 3 seconds (or pressing the dedicated USB management button if one is present)
2. Connect the mobile device via a USB cable to the server's USB port with the management symbol
3. In iOS or Android settings, enable Personal Hotspot or USB Tethering
4. Launch the Lenovo XClarity Mobile app

Remote management

The server offers a dedicated RJ45 port at the rear of the server for remote management via the XClarity Controller management processor. The port supports 10/100/1000 Mbps speeds.

IPMI via the Ethernet port (IPMI over LAN) is supported, however it is disabled by default. For CTO orders you can specify whether you want to the feature enabled or disabled in the factory, using the feature codes listed in the following table.

The SR860 V2 has XCC Enterprise as a standard feature. XCC Enterprise includes a virtual presence (remote control and remote media) capability for remote server management.

The remote control functions include the following:

• Remotely viewing video with graphics resolutions up to 1600x1200 at 75 Hz with up to 23 bits per pixel, regardless of the system state
• Remotely accessing the server using the keyboard and mouse from a remote client
• Capturing blue-screen errors
• International keyboard mapping support
• LDAP-based authentication
• Boot Capture
• Remote mounting of ISO and IMG files
• Virtual console collaboration - Ability for up to 6 remote users to be log into the remote session simultaneously
• Power capping
• License for XClarity Energy Manager

Because Lenovo XClarity Administrator does not require any agent software to be installed on the managed endpoints, there are no CPU cycles spent on agent execution, and no memory is used, which means that up to 1GB of RAM and 1 - 2% CPU usage is saved, compared to a typical managed system where an agent is required.

Lenovo XClarity Administrator is an optional software component for the SR860 V2. The software can be downloaded and used at no charge to discover and monitor the SR860 V2 and to manage firmware upgrades.

If software support is required for Lenovo XClarity Administrator, or premium features such as configuration management and operating system deployment are required, Lenovo XClarity Pro software subscription should be ordered. Lenovo XClarity Pro is licensed on a per managed system basis, that is, each managed Lenovo system requires a license.

The following table lists the Lenovo XClarity software license options.

For more information, refer to the Lenovo XClarity Administrator Product Guide:
http://lenovopress.com/tips1200

For more information about all the available Lenovo XClarity Integrators, see the Lenovo XClarity Administrator Product Guide: https://lenovopress.com/tips1200-lenovo-xclarity-administrator