HPE Apollo is HPE's line for workloads that outgrow a standard ProLiant rack server: extreme storage density, scale-out computing and multi-GPU AI. Apollo is not one product but several families, each built for a single job. The short answer: the Apollo 4200 for storage-heavy work such as Hadoop, Ceph and object storage; the Apollo 2000 for many compute nodes in little rack space; and the Apollo 6500 for multi-GPU AI training. Pick by workload, not by name.
| Family | Role | Headline figure (as listed by Serverwale) | Best for |
|---|---|---|---|
| Apollo 4200 / 4000 | Storage-dense | Up to 28 LFF + 6 SFF drives, about 392 TB in 2U (4200 Gen10) | Hadoop, Ceph, object and cold storage |
| Apollo 2000 | Multi-node density | Up to 4 dual-socket nodes in 2U, shared power and cooling | Scale-out HPC, VDI, web farms |
| Apollo 6500 | GPU / AI compute | Up to 8 double-wide GPUs with NVLink topology (Gen10) | Deep-learning training, GPU HPC |
The figures come from Serverwale's HPE Apollo servers page. Availability, generation (Gen9 or Gen10) and exact configuration depend on the stock at the time.
The Apollo 4200 packs a large number of large-form-factor drives into 2U. That suits data that is big and mostly read in bulk: data lakes, Ceph or MinIO object storage, and backup targets. Because so many drives sit in one chassis, protection matters: read RAID levels explained and remember that RAID is not a backup. If you are deciding between a storage server and a shared array, our SAN vs NAS guide helps.
The Apollo 2000 puts up to four dual-socket nodes into a 2U chassis that shares power supplies and cooling. This lowers cost per node and rack space for clusters of identical machines. The trade-off is shared infrastructure: the nodes depend on the same chassis, so plan for that in your redundancy and spares. See PSU redundancy and multi-node vs scale-up servers.
The Apollo 6500 Gen10 is built to host up to eight double-wide GPUs, and Serverwale lists NVIDIA A100, V100 and RTX options. For AI work, GPU memory is the first sizing question: our GPU VRAM sizing guide shows how to work it out. For a short project, the Serverwale blog compares renting vs buying a GPU server; for smaller builds, see GPU servers.
Dense systems put more heat and power draw into each rack unit, so check that your room, PDUs and UPS can support them before ordering. Our guides on rack PDU power, rack UPS sizing and cooling and airflow cover the checks.
According to its Apollo page, Serverwale sells refurbished Apollo systems at 60–80% below new, iLO-managed, with a documented 40-point test, firmware updates and burn-in, a 3-year warranty and GST invoice. The page also says Apollo systems are available on monthly rental, with the option to convert to purchase; confirm the terms in your quote. For a standard rack workload an HPE ProLiant is usually enough; compare in our Dell R740 vs HPE DL380 guide.
What is HPE Apollo used for? High-performance computing, AI training, big-data analytics, object and cold storage, VDI and large web farms. The platform trades general-purpose flexibility for density in compute, GPU or storage.
Which Apollo model is best for AI? The Apollo 6500, which supports up to eight double-wide GPUs with NVLink topology on Gen10.
Which Apollo model is best for storage? The Apollo 4200 or 4000 families, which hold the most drives per rack unit; the 4200 Gen10 is listed at up to 28 LFF + 6 SFF drives in 2U.
Can I rent an HPE Apollo server? Serverwale lists Apollo systems for both refurbished purchase and monthly rental. Confirm the terms and any option to convert to purchase in your quote.
Is refurbished Apollo reliable for production HPC? Serverwale states each unit passes a documented 40-point test with firmware update and burn-in and carries a 3-year warranty. Ask for the test details of the specific unit.
Tell us your workload and we will suggest the right Apollo family and configuration. Serverwale · Phone +91-87962-44410 · WhatsApp https://wa.me/918796244410 · Burari, Delhi 110084, India. You can also browse live stock.
Prefer a brand-new build? ProStation Systems custom servers are built to order.