MongoDB on Lenovo ThinkAgile HX650 V4 and FX650 V4

Modern applications increasingly generate dynamic, semi-structured data: customer profiles, product catalogs, clickstream events, IoT telemetry, application logs, and operational metrics. These data sources rarely fit neatly into rigid relational schemas, and organizations are turning to document-based databases that evolve alongside the applications they support. MongoDB has become a leading operational database for these scenarios, combining a flexible document model with horizontal scalability, built-in replication, and native support for vector and full-text search that powers retrieval-augmented generation (RAG) and other AI applications.

At the same time, infrastructure teams are consolidating compute, storage, and virtualization onto hyperconverged platforms to simplify operations and reduce the number of systems they manage. The arrival of Intel Xeon 6 processors (code named "Granite Rapids") in the Lenovo ThinkAgile HX650 V4 and FX650 V4 gives that consolidated infrastructure a substantial increase in core count and memory capacity, precisely the resources MongoDB needs to handle insert-heavy and mixed read/write workloads at scale.

Figure 1. ThinkAgile HX650 V4 (top) and HX630 V4 (bottom) for Nutanix hyperconverged infrastructure

Choosing infrastructure for MongoDB is rarely just a sizing exercise. IT teams evaluating a platform for document databases face several intersecting pressures:

• Unpredictable performance at scale: as document sizes and concurrent client counts grow, throughput and latency do not scale linearly, making it hard to size infrastructure with confidence for insert-heavy ingestion or mixed transactional workloads.
• Fragmented AI infrastructure: adding retrieval-augmented generation and vector search often means standing up a separate vector database and search engine, increasing operational complexity and latency between systems.
• Siloed compute and storage: traditional three-tier architectures separate compute, storage, and virtualization, adding management overhead and slowing the ability to scale a MongoDB cluster on demand.
• Locked-in hypervisor choices: many hyperconverged platforms tie an organization to a single hypervisor, limiting flexibility as licensing, cost, and operational requirements change over time.
• Lifecycle and data protection burden: provisioning, cloning, backing up, and patching database instances consistently across a growing MongoDB estate consumes scarce administrator time.

These pressures push organizations to look for infrastructure that is validated for MongoDB, flexible in its software stack, and simple to operate at scale.

Lenovo addresses these challenges with the ThinkAgile HX650 V4 and FX650 V4 hyperconverged systems, 2-socket, 2U platforms (also available as 1U ThinkAgile HX630 V4 and FX630 V4 systems) built on Intel Xeon 6 processors. Both systems are fully validated and integrated by Lenovo, certified and preloaded with licensed Nutanix software, and designed to run the Nutanix Cloud Platform (NCP). NCP unifies compute, storage, and virtualization into a single hyperconverged layer that provides enterprise storage, integrated data protection, built-in resilience, unified management, advanced analytics, and end-to-end security for mission-critical workloads such as MongoDB.

Figure 2. ThinkAgile FX650 V4 (top) and FX630 V4 (bottom) designed for flexible hyperconverged infrastructure

ThinkAgile FX650 V4 adds a further layer of flexibility: organizations can move between hypervisors of their choice while keeping the same integrated compute, storage, and management stack, future-proofing their investment against hypervisor licensing or strategy changes.

MongoDB runs as a virtual machine on the Nutanix AHV hypervisor, using the WiredTiger storage engine, and can take full advantage of the high core counts and large memory capacity that Intel Xeon 6 processors and dense NVMe storage provide. For teams that want to standardize database operations, Nutanix Database Service (NDB) registers, provisions, clones, and administers MongoDB databases across one or more Nutanix clusters in single node or multi-node configurations; MongoDB can also be deployed and scaled manually outside of NDB. Lenovo XClarity One provides ongoing systems management across the cluster, backed by Lenovo's broad selection of processor, memory, and drive options to match the deployment to the workload.

Lenovo ThinkAgile HX650 V4 and FX650 V4 support MongoDB across a range of enterprise scenarios, from consolidating private cloud infrastructure to running the ingest-heavy and AI-driven workloads that modern applications generate. The following use cases show how organizations put the platform's performance, flexibility, and validated benchmark results to work.

• Private cloud consolidation and hypervisor flexibility

  Enterprises consolidating MongoDB alongside other virtualized workloads onto a single private cloud platform deploy ThinkAgile HX650 V4 or FX650 V4 running the Nutanix Cloud Platform. The hyperconverged architecture combines compute, storage, and virtualization into one manageable system, reducing the number of silos administrators must maintain. Organizations that want the option to change hypervisors over time choose ThinkAgile FX650 V4, which lets them move between hypervisors of their choice without re-architecting the underlying compute, storage, or management stack, protecting their MongoDB investment as strategy or licensing needs evolve.

• Mixed read/write transactional applications

  Applications such as order processing, customer profile updates, and catalog management generate a steady mix of reads and writes against MongoDB. Benchmark testing on ThinkAgile HX650 V4 showed the platform sustaining strong throughput across both balanced 50/50 and read-heavy 80/20 read/write workloads, with the 80/20 profile remaining especially resilient even as document size grew to 8,192 bytes. For teams running transactional, customer-facing MongoDB applications, this means predictable performance across the document sizes and concurrency levels typical of production traffic.

• High-volume operational data

  IoT telemetry, clickstream events, application logs, and operational metrics arrive continuously and at high volume, making insert throughput the critical metric. Testing showed ThinkAgile HX650 V4 sustaining over 100,000 load operations per second for small and medium-size documents (256 bytes and 2,000 bytes) across 64 to 128 concurrent client threads. The platform's high core-count Intel Xeon 6 processors and NVMe-based local storage let MongoDB absorb ingest-heavy traffic from streaming analytics, IoT platforms, and other large-scale operational systems without sacrificing responsiveness.

• AI-ready search and retrieval-augmented generation

  MongoDB's native support for vector search combined with full-text search lets organizations retrieve the most relevant results to power AI applications and agentic systems that use retrieval-augmented generation (RAG). Running this capability on ThinkAgile HX650 V4 or FX650 V4 keeps vector, text, and operational data together in the same self-managed, on-premises MongoDB deployment, eliminating the need for a separate external vector database or search engine and reducing the complexity and latency of building RAG pipelines directly where enterprise data already lives.

ThinkAgile HX650 V4 and FX650 V4 are validated for the workload types that make up most production MongoDB deployments, from high-volume data ingestion to AI-driven search. Rather than testing a single synthetic benchmark point, Lenovo varied document size, client concurrency, and read/write mix to mirror the range of traffic patterns these workloads actually generate. The table below summarizes the representative workload profiles and how each one performed.

Across every workload profile tested, ThinkAgile HX650 V4 sustained strong throughput with consistently low latency, giving IT teams a proven performance envelope to size against rather than a single best-case number. Full lab configuration and measured results follow below.

Lenovo validated this configuration on a four-node Nutanix cluster built from Lenovo ThinkAgile HX650 V4 systems, with MongoDB deployed as a single, standalone instance, not a replica set or shared cluster, inside one large virtual machine sized to take advantage of the platform's core count and memory capacity. Across every workload tested, CPU utilization stayed below 50% while memory was the more constrained resource, so production deployments should size CPU and memory independently to match their own workload rather than treating this lab configuration as a fixed template.

The following table lists the configuration of each ThinkAgile HX650 V4 host.

The following table lists the MongoDB virtual machine configuration.

Testing used the Yahoo Cloud Serving Benchmark (YCSB) against a single, standalone MongoDB instance, using a 10 million record, 10 million operation dataset across three document sizes and three client concurrency levels.

The following table shows the measured throughput for every combination tested.

The following table shows the average latency for every combination tested.

Across all three workload profiles, ThinkAgile HX650 V4 delivered strong throughput and stable latency. Peak throughput reached 116,149 operations per second on the 80/20 read/write workload with 256-byte documents at 64 threads, and load throughput exceeded 100,000 operations per second for both 256-byte and 2,000-byte documents. Latency stayed low and predictable for small and medium documents, typically under 1 millisecond at 64 and 96 threads.

Larger, 8,192-byte documents showed the expected trade-off: the balanced 50/50 workload saw average update latency rise to 2.484 ms at 96 threads, while the read-heavy 80/20 workload stayed far more resilient, with both read and update latency noticeably lower at the same document size and thread count. That resilience means organizations running read-dominant MongoDB applications can expect the platform to hold its performance even as payloads grow, while write-heavy applications benefit most from keeping documents compact.

Testing also showed these workloads to be memory-bound rather than CPU-bound: CPU utilization stayed below 50% throughout, even at the highest concurrency levels, on a standalone MongoDB instance rather than a replica set or shared cluster. Organizations planning a production deployment should size CPUs and memory independently to match their own workload profile, rather than treating this lab configuration as a fixed template.

The feature codes below are the exact configuration used for this validation, for one Lenovo ThinkAgile HX650 V4 server node; multiply quantities by four for the full cluster. Use them as a starting point when configuring an order with your Lenovo representative or Lenovo Business Partner.

MongoDB running on Lenovo ThinkAgile HX650 V4 and FX650 V4 with Nutanix software delivers strong throughput and consistently low latency across the document sizes, concurrency levels, and workload mixes that enterprise applications actually generate. The platform combines validated, integrated hardware and firmware with licensed Nutanix Cloud Platform software, giving organizations a single hyperconverged foundation for private cloud, AI/ML, and analytics workloads built on MongoDB. Whether consolidating infrastructure, scaling operational data ingestion, running mixed transactional applications, or building retrieval-augmented generation pipelines, ThinkAgile HX650 V4 and FX650 V4 give MongoDB the scalable, predictable performance modern data platforms require.

To learn more about MongoDB on Lenovo ThinkAgile HX650 V4 / FX650 V4, contact your Lenovo representative or Lenovo Business Partner, or visit the resources below.

References:

• Running MongoDB on Lenovo ThinkAgile HX V4 and FX V4 Servers with Intel Xeon 6 Processors (LP2412):
  https://lenovopress.lenovo.com/LP2412
• Lenovo ThinkAgile HX Series:
  https://www.lenovo.com/us/en/servers-storage/thinkagile/hx-series/
• Lenovo ThinkAgile FX Series:
  https://www.lenovo.com/us/en/servers-storage/thinkagile/fx-series/
• MongoDB:
  https://www.mongodb.com/
• Nutanix Cloud Platform:
  https://www.nutanix.com/products/cloud-platform
• Multi-Vendor Hyperconverged Infrastructure with Lenovo ThinkAgile FX V4 (LP2520):
  https://lenovopress.lenovo.com/lp2520-multi-vendor-hyperconverged-infrastructure-with-lenovo-thinkagile-fx-v4

Chandrakandh Mouleeswaran is a Solution Architect with 18+ years of experience in software development, performance testing and engineering, having worked on designing and architecting many scalable enterprise applications. He has spent a decade in technical enablement and partner solution development for VMware, Nutanix, Oracle and other ISVs across industries and technologies. He specializes in architecting infrastructure solutions for virtualization, VDI, database, cloud, data science, AI/ML solutions and various enterprise workloads.

Cristian Ghetau is an Advisory Engineer for Lenovo in Romania and has experience in Cloud Infrastructure technologies. He has had more than 13 years of experience working with virtual environments from VMware, Microsoft, Oracle, and Linux.

Chris Honoré is a Solutions Product Manager at Lenovo with deep expertise in datacenter products and solution offerings. He has a strong background in consulting and solution development, helping customers design and support on-premises and hybrid environments. Chris has spent the past 15 years with IBM and Lenovo, specializing in x86 server and data center solutions. Prior to that, he built two decades of experience in the telecommunications industry, serving in both technical and business leadership roles.