Migrating from HPE ProLiant ML350 Gen9 to ML350 Gen11 — Parts, Timeline and Budget Guide

Written for IT leaders planning an HPE ProLiant refresh in 2026. All specs, EOSL dates, and part numbers come from HPE QuickSpecs and ICD’s live catalog. Edge / branch / lab environments that ran Gen9 towers for a decade are the classic targets here. Gen11 ML350 tower chassis is larger but quieter and handles 4TB DDR5.

Why this migration

ML350 Gen9 reached HPE End of Sale on 2020-07-22 and End of Service Life on 2025-07-22. Beyond EOSL, HPE post-warranty pricing escalates, SPP firmware bundle cadence drops, and iLO CVE patches tail off. ML350 Gen11 resets the five-to-seven-year lifecycle clock, brings DDR5 bandwidth, AMX AI tile instructions, and — critically in 2026 — aligns you with HPE’s active security train (iLO 6).

OEM support timeline (real HPE EOSL dates)

Your ML350 Gen9 has HPE End of Service Life on 2025-07-22. That means HPE parts availability becomes best-effort, OneView firmware bundles ship less frequently, and critical CVE patches for iLO/BIOS taper off.

Your ML350 Gen11 sits inside HPE’s active lifecycle — roadmap-committed through at least 2030 for Gen11 and 2031+ for Gen12. New HPE SPP (Service Pack for ProLiant) drops land first on the active platforms, which means your security posture improves measurably post-migration.

Compatibility analysis

The DDR4 → DDR5 transition is the single biggest compatibility barrier. DDR5 UDIMM/RDIMM uses 288 pins at 1.1V versus DDR4 at 1.2V; notch positions differ; PMIC is on-DIMM rather than on-motherboard. Zero mechanical carry-over. Expect full memory replacement for this path.

Smart Array → MR controller implications

The controller line change from HPE Smart Array (HPE-designed ASIC) to Broadcom MegaRAID (MR) on Gen11 is more than a rebrand. Three operational impacts matter in a migration:

1. Cache protection hardware changes. Smart Storage Battery (871264-001) is not compatible with MR controllers. You need a Smart Storage Hybrid Capacitor (P01367-B21 typical) sourced new.
2. Array metadata format changes. You cannot physically move a Smart Array P840ar-flashed array into an MR216i-p and expect it to import. The data must be migrated via replication (storage vMotion, DB log shipping, rsync, etc.) — not drive-pulled and re-inserted.
3. Management tooling changes. HPE Smart Storage Administrator (SSA) scripts and Ansible modules do not target MR controllers. Rewrite or port to storcli64 / Redfish MR schemas.

Data migration path

Depending on what sits on the ML350 Gen9, the data path varies. For hypervisor hosts (VMware vSphere, Nutanix, KVM, Hyper-V) the cleanest path is live workload migration — vMotion, Nutanix Move, Live Migration — node-by-node. Provision one ML350 Gen11 alongside the cluster, add it as a node, evacuate one ML350 Gen9, retire the ML350 Gen9, repeat. ICD has delivered 40+ of these phased VMware refreshes in the region with near-zero downtime.

For bare-metal databases (Oracle, MSSQL, PostgreSQL) use native replication: Oracle Data Guard, SQL Server AG, PG streaming replication. Target ML350 Gen11 becomes the standby, then failover during a planned window. This protects RPO strictly and lets you roll back if the ML350 Gen11 hits an unexpected issue during burn-in.

For file servers, backup targets, and archive the classic rsync or Robocopy approach works — but plan for longer windows if you are moving from 12G SAS SFF to NVMe E3.S, as the target can ingest 4-10x faster than the source can read.

Parts shopping list (real PNs in ICD’s catalog)

• Target CPU (ML350 Gen11): P49617-B21 (Xeon 3408U G11), P49610-B21 (Xeon 4410Y G11), P49597-B21 (Xeon 5415+ G11), P49611-B21 (Xeon 4416+ G11)
• Target memory (DDR5): P43328-B21 (32GB RDIMM DDR5-4800), P43331-B21 (64GB RDIMM DDR5-4800), P50311-B21 (32GB 2Rx8 DDR5-4800), P50312-B21 (64GB 2Rx4 DDR5-4800), P43334-B21 (128GB DDR5-4800 3DS)
• Target controller: P26324-B21 (MR216i-a 12G), P26279-B21 (MR416i-a 12G), P06367-B21 (MR416i-p 12G), P04220-B21 (SR416i-a 24G Tri-Mode), P12688-B21 (SR416i-a 24G)
• Target PSU: P38995-B21 (800W FS Platinum), P18224-B21 (800W FS Titanium), P67240-B21 (1000W m-CRPS Titanium), P73190-B21 (800W m-CRPS Titanium)
• Source spares available (ML350 Gen9) for ICD Care+ bridge: 826846-B21 (Xeon Silver 4110), 826850-B21 (Xeon Silver 4114), 726718-B21 (8GB RDIMM DDR4-2133), 774172-001 (16GB DDR4-2400R), 726736-B21 (P440ar / 2GB FBWC), 749974-B21 (P840 / 4GB FBWC)

Migration phases and next steps

1. Baseline (week 1). Export current iLO configs, HPE OneView inventory, Smart Array drive maps, BIOS settings, and firmware levels for every ML350 Gen9 node in scope. ICD provides a free export script on request.
2. BOM design (week 1-2). RFQ to ICD for BTO ML350 Gen11 configs. ICD returns a cost-audited BOM in 24 hours with HPE part numbers, warranty posture, and lead-time visibility.
3. Pilot (week 3-5). Deploy one ML350 Gen11 pilot node. Migrate one non-critical workload (typically a dev/test VM, a secondary DB replica, or a file-server target). Validate performance under load, iLO Redfish scripts, and your backup integration.
4. Rolling production cutover (week 5-12). Phased migration by application window. Each ML350 Gen9 node evacuated, retired, optionally repurposed for DR or returned to ICD under buy-back credit.
5. Post-migration (week 13+). Turn off HPE support on the retired fleet (or keep it if you are holding as cold spares), re-baseline capacity plans against the new platform headroom, and document the new BOM for next-cycle planning.