I am looking at storage refresh within the next 6 months. I have been in touch with my EMC sales rep, but I wanted to hear from EMC's customers that have been on their box a bit. I currently have VMAX 10K single engine that runs 150 servers and about 900 desktops pretty good. I started looking at the price of adding more drives(4 DAES), but the box is about 3 years already and there are more current boxes out there. I have been doing some reading on XTremeIO and have several discussions with EMC about it. I have the Dedupe analysis results and it states it can dedupe it very well. Can/Should I mix the View and Server workloads on this like I currently do a single Engine VMAX? Should we get multiple bricks, same cluster or several single brick clusters? I know a lot of this depends on price, which I do have.
Nearly everything on the SAN is VMware,
But my real questions,
Is the interface too cluttered or dumbed down with too many wizards?
Are there Java issues to worry about it to compatibility reasons? I wish EMC would have a non JAVA GUI, I am worried that Java support in the browsers is rapidly disappearing. Chrome no longer has it, Firefox is dropping it, Edge doesn't have it, who knows with Safari and IE.
How have the upgrades been?
Any issues or good things have anyone had with their box? even comparisons to other boxes will be nice.
Maybe I was spoiled with the other AFA POCs but the other AFA did not require all the cabling, less FC ports and actually had a smaller rack footprint than XIO did. Check the back of XIO that is racked and up, looks like a sphagetti bowl.
Not a deal breaker but longer set up times and maybe not as easy to troubleshoot connectvity issues. I think if EMC designs a chassis that can fit all the components of an Xbrick while being modular it would simplify the setup, expansions, upgrades and connectivity troubleshooting.
Some bugs restarted the XIO cluster automatically and brought us to halt during production hours. Needless to say it is not going back into production anytime soon.
It was a rare confluence of bugs and probably never will happen again but it did happen so I cannot tell the business it is reliable for production yet.
Thank you for your feedback. Allow me to comment on a few things that you have brought up.
We have come up with a cable management arm for XtremIO which will make the cables much more manageable on the back of an XtremIO array. Having said that, XtremIO is a scale out architecture with more components (more number of storage controllers, as an example) - so it is going to have more cables compared to an array which is based on a dual controller scale-up architecture. But the benefit is in the linear scalability, consistent and predictable performance that XtremIO can provide as you scale out in terms of the number of X-Bricks.
As for the other bugs that you ran into, we can help provide an ETA of those bug fixes if you dont have that already. Feel free to message me privately, since this is not the right forum to be discussing customer specific details. I hope you understand.
Thank you for your feedback. Allow me to comment on a few things that you have brought up.
We have come up with a cable management arm for XtremIO which will make the cables much more manageable on the back of an XtremIO array. Having said that, XtremIO is a scale out architecture with more components (more number of storage controllers, as an example) - so it is going to have more cables compared to an array which is based on a dual controller scale-up architecture. But the benefit is in the linear scalability, consistent and predictable performance that XtremIO can provide as you scale out in terms of the number of X-Bricks.
Avi, scale out is your excuse for the rats nest i see behing my XtremIO box ? Talk to VMAX guys, they can tell you how to do scale out with prestine cable management that spans not only floor tiles but aisles.
It is not an excuse by any means. We do appreciate any and every feedback that comes in from our customers.
As I mentioned earlier, we do recognize that there is a scope for improvement and have come out with the cable management arm for new arrays that are shipping out now.
Being a new XtremIO customer with multiple vendors in the data centers. Some things to improve on reducing mess at the back:
1. Service Controllers, DAE can be combined together in one 4U box and cabling done internally/backplane. Reduces few moving parts behind the back.
2. Can we eliminate the BBUs and manage power failures differently ? That reduces some more cabling.
3. Can you ship right length cables for the customers so everything is neat and tight
Each time customer/vendor touches a cable behind the back for troubleshooting, it takes couple of hours to make sure system health is normal due to the cabling mess at the back.
blubud
1 Rookie
•
35 Posts
1021
0
Posted November 17th, 2015 12:00
In a nutshell so far...
1. Performance and scalabilty is good.
2. Interface and the array is really easy to use once everything is setup.
2. Reliability poor. The OS is buggy needs to mature. We pulled it from production.
3. The hardware needs to be simplified.
4. No NVMe so hardware upgrades are needed in the future to keep up with the maturing SSD landscape.