all systems operational
Get started
128 GB DDR4 ECC · registered

128 GB RAM dedicated servers

128 GB DDR4 ECC memory across two platforms — 2× Xeon E5-2690 v3 (Haswell, 24c) or 2× Xeon Gold 6138 (Skylake-SP, 40c). Both from $500/mo.

128 GB
DDR4 ECC registered
4 channels
per socket
~68 GB/s
memory bandwidth
from $500
10 Gbit unmetered

About 128 GB RAM dedicated server

128 GB is the workhorse RAM tier — enough for a decent MySQL / Postgres master (~10-40 M row tables in buffer pool), a 20-VM KVM host, a mid-size Redis cluster, or a well-caches CDN edge node.

Both our 128 GB SKUs use registered DDR4 ECC RDIMMs at DDR4-2133 (E5-2690 v3) or DDR4-2666 (Gold 6138). ECC catches single-bit memory errors — matters for long-running processes and financial workloads. Memory topology: 4 channels per socket → ~68 GB/s aggregate bandwidth.

Typical customers on 128 GB tier: WordPress fleets (5-15k daily users per site), Magento shops during flash sales, mid-scale IPTV panels with in-memory catch-up cache, Elasticsearch clusters, Redis-heavy application caches, JVM services with large heaps.

Both platforms support up to 512 GB (E5-2690 v3) / 1.5 TB (Gold 6138) memory expansion. Upgrading to 256/384/512 GB is a hardware swap — 30-60 min downtime, we handle it.

/faq

Frequently asked questions

What's the max RAM I can add later?

E5-2690 v3 → 512 GB (16 DIMM slots × 32 GB). Gold 6138 → 1.5 TB (24 DIMM slots × 64 GB). We handle the upgrade in a scheduled maintenance window.

Is ECC memory really necessary?

For anything running 24/7 with critical data — yes. Single-bit errors happen once per few TB of RAM per year statistically. ECC transparently corrects them. Non-ECC in a database server is a bad idea.

Registered vs unbuffered DIMMs?

All our 128+ GB servers use registered (RDIMM) or LRDIMM. Required by Xeon E5/Gold platforms above 64 GB per socket. Slightly higher latency, much better scalability.

Chat with us@hostfory