Last week, thanks to Tech Field Day Extra, I attended a presentation from the EMC’s XtremIO team. Some of my concerns about this array are still there but there is no doubt that this product is maturing very quickly and enhancements are released almost on a monthly basis… and it’s clear that it has something to say.
A rant about All Flash
In these days, contrary to the general (and Gartner?) thinking, I’m developing the idea that considering All Flash Arrays a separate category is a totally non sense (you can also find an interesting post from Chris Evans about this topic). Flash memory is only a media and storage should be always categorized looking at its characteristics, features and functionalities. For example, I could build a USB-keys based array at home, it’s AFA after all… but would you dare saving your primary data into it? Will it be fast? (you don’t have to answer, of course!)
The fact that a vendor uses Flash, Disks, RAM or a combination of them to deliver its promises is only a consequence of designing choices and we have to look at the architecture (both hardware/software) as a whole to understand its real world positioning. Resiliency, availability, data services, performance, scalability, power consumption and so on, are the characteristics you still have to consider to evaluate if an array is good for a job or another.
Back to XtremIO
In this particular case, If we go back and look deeply into XtremIO design we will find that the system is equipped with plenty of RAM which is heavily leveraged to grant better constant performance and the highest predictability. In fact, looking at the charts shown during the presentation (around minute 14 of the video below), you’ll find that the system, no matter the workload, delivers constant latency well under the 1ms barrier.
The product, which has finally received updates enabling all common data services expected on a modern storage array (replication is still missing though), doesn’t shine for power consumption, used rack space or other kinds of efficiencies (at this time it’s also impossibile to mix different type of disks for example). But again, granting first class performance and predictability is always the result of a give-and-take.
XtremIO is based on a scale-out architecture with a redundant infiniband backend. Different configurations are available starting from a single brick (a dual controller system and its tray populated with 12 eMLC drives, out of the 25 available) up to a six-brick configuration for a total of 90TB (usable capacity before deduplication/compression). No one gave me prices… but you know, if you ask the price you can’t afford it (and, of course, they are very careful to that because $/GB really depends on the size of the array and deduplication ratio you can obtain from your data).
Why it is important
XtremIO is strongly focused on performance and on how it’s delivered. From this point of view it clearly targets traditional enterprise tier 1 applications and it can be considered a good competitor in that space. It clearly needs some improvements here and there but EMC is showing all its power with the impressive quantity of enhancements that are continuously added.
You know what? From my point of view, the worst part of EMC XtremIO story is that there isn’t a simple and transparent migration path from the VMAX/VNX, which would be of great help for the end user (and EMC salesforce)…
Other videos from EMC XtremIO session:
Disclaimer: I was invited to Tech Field Day Extra by GestaltIT and they paid for travel and accommodation. VMworld Pass was kindly provided by VMware and many vendors will have special giveaways for bloggers and vExperts.
I have not been compensated for my time and am not obliged to blog. Furthermore, the content is not reviewed, approved or published by any other person than the Juku team.