What is most valuable?
Synchronous replication. One group, one pool, one IP address, the member stands in two different datacenters with, for example, two hosts in DC1 and two hosts in DC2 in same ESX cluster. If you have one datacenter fail, the VMs will start to run inside the other data center.
How has it helped my organization?
EQL is pure block storage. EQL has not improved anything. It created a rigid change regime and bureaucracy after several firmware upgrade failures.
What needs improvement?
Block size, controllers, failover mechanism in general for the controllers. 6510 and 6610 have two iSCSI connections in each controller. Why do 6110xxx members have only one, and why do the same members have the possibility of having an active iSCSI interface in the passive controller? It could make a lot of noise with a failover.
For how long have I used the solution?
What do I think about the stability of the solution?
This product was very unstable in the beginning. A pool with more than one member could get crazy and start to empty one member to another member (if the pool went below 10% free capacity [from Dell], for real EQL processes started between 11-12% capacity). One member became full and crashed production which used the EQL pool.
Fixed somewhere in v.6.1.x!? Two 6510es in pool generate, for example, 125TB. So we need 10% free capacity, 12.5TB in the pool to make sure EQL does not go crazy.
What do I think about the scalability of the solution?
Number of iSCSI connections. The EQL group has a maximum of four pools (at least in v.7xxx). You need at least one pool as a "maintenance" pool. In practical terms, we have only three useful pools in one group.
How are customer service and technical support?
Horrible. Very rarely do you meet people with a passion for the product. Meaning, how can people have passion for this product series?
Which solution did I use previously and why did I switch?
Started with 14x 6110x to run OS and regular data and some SQL. Filled up the IO long before capacity. Split up the groups and reformatted some of the EQLs to RAID 10 to run only the OS. That worked well. IO and storage capacity was quite even. Dell was very positive against snapshots. They forgot about their 15-16MB block size. Local snapshots on data stores which only run OS which generates 20% random reads and around 80% writes.
Data retention was one snap each day in a seven day rotation. We got around 150% on the top of the datastore in only snapshot storage. EQL cheap? Hmmm... I don't think so.
How was the initial setup?
What's my experience with pricing, setup cost, and licensing?
All features are out of the box. The license will be in the virtual, OS, or application layer.
What other advice do I have?
Do not go for EQL. It will be out of production in a few years. Dell EMC Unity is a much better option for price, performance, and safety.
If you don't have alternatives, management should be fired! But seriously, you need to have very good control of your workload and IOPS. Which data cause random reads, sequential reads, random writes, sequential writes, etc. And you need to build your environment. The workload needs to be separated in VMs on different disks which touch different data stores and SAN environments.
Our relationship with Dell EMC is very good. Dell has always been a good hardware vendor and are really good at it. They are also good in local storage in, for example, their MD series. Really good!
They could not make the SAN part in EQL and Compellent. Maybe that was the reason Dell bought EMC.
We are also using Dell F10 which provides quite good performance. Hardware is the same as Cisco, but there are some downs with their software. F10 doesn't have any control over some packages, such as pause frames. The switch will try to send those packages to all ports. So if you have activate flow control, you need to turn off RX or TX or the switch will flood all the ports with "unknown" packages.
In the end. We will replace all EQLs with Dell EMC Unity. We have not upgraded EQL after 7.1.8, which is quite stable, because firmware upgrades cause us alot of trouble.