Enterprise storage: all-flash arrays, SAN and NAS

Enterprise storage separates the disks from the servers so capacity and performance can be managed once instead of per machine. The practical split is by access protocol: block (SAN) for databases and virtual machines, file (NAS) for shared documents and media, object for archive and backup at scale. We supply entry SAN through tier-1 all-flash and scale-out platforms.

Product families we supply

Product familyBrandCapacity / formatProduct page
OceanStor Dorado (all-flash)Huaweientry to high-end NVMe
OceanStor PacificHuaweidistributed / scale-out
OceanProtectHuaweibackup and archive
PowerStore (unified)Dellblock + file + vVol, NVMeView product
PowerVault ME5Dell2U/5U; up to 8 PB
PowerScale (NAS)Dellscale-out NAS
PowerMaxDellenterprise high-end
MSA Gen6 (giriş SAN)HPE2U; up to 7.37 PB; 395,000 IOPS
Alletra Storage MPHPEblock and object, disaggregated architecture
Alletra 5000 / 6000HPEmid-range
AFF A-Serisi (yeni nesil)NetApp40M IOPS, 1 TB/s, 99.9999% availabilityView product
AFF A-Serisi (önceki)NetAppall-flash NVMe
ASA (SAN’a özel)NetAppblock storage
FAS (hibrit)NetApphybrid disk + flash
E-SerisiNetApphigh bandwidth / HPC
FlashArray//XPure Storagetier-1 block, NVMe DirectFlashView product
FlashArray//XLPure Storagehighest performance
FlashArray//CPure Storagecapacity-oriented QLC
FlashArray//E ve FlashBladePure Storageobject and file; low-cost capacity

Product families we actually source. Capacities are the manufacturer’s stated series ranges, not a quotation. Exact order codes depend on voltage, outlet type, battery configuration and region and are confirmed on the quote.

Which one should you choose?

Block, file or object?

Block storage presents raw volumes and is what a hypervisor or database wants. File storage presents shares and is what people and applications want. Object storage is addressed over HTTP and is where backups and archives belong. Unified arrays do the first two in one box, which is usually the right answer below a few hundred terabytes.

All-flash or hybrid?

All-flash has become the default for anything transactional; the price gap has narrowed and the power and rack-space saving is real. Hybrid still makes sense where the working set is small and the bulk is cold, but check the tiering behaviour, because a hybrid array with the wrong working set performs like its slowest tier.

How do you size it?

Usable capacity, not raw. Take the raw figure, subtract RAID or erasure-coding overhead, subtract spares, then decide whether you trust the vendor’s data-reduction ratio. Ask for both the guaranteed effective capacity and the raw figure. The difference between the two is the part of the quote worth arguing about.

Performance: IOPS or throughput?

Databases and virtual desktops are IOPS and latency problems. Backup, video and analytics are throughput problems. An array tuned for one is often mediocre at the other, so give us the workload rather than a target IOPS number copied from a datasheet.

Growth: scale-up or scale-out?

Scale-up adds shelves to one controller pair and is simpler and cheaper until the controllers saturate. Scale-out adds nodes that bring their own controllers, so capacity and performance grow together. The right model for file and object once you are past a few hundred terabytes.

Replication and backup are separate purchases

Array-based replication protects against a site failure, not against deletion or ransomware. Snapshots help but live in the same system. If backup is in scope, say so at the quote stage: the target and the licence are separate line items and are commonly forgotten until the array is already installed.

Request a quote

Tell us usable capacity, the protocol you need, the workload and the delivery country. If replication or backup is in scope, mention it: those are separate line items.

Request a Quote (EN)