When FAST is not properly fast.

I have just finished read this, this and this about the “brand new”, next to be released, FAST feature from EMC on their V-MAX. In his post Barry Burke the Anarchist asks us to place some questions.. so here are my questions and thoughts!

FAST (fully automated storage tiering) is an implementation of what’s commonly called Automated Tiered Storage, this feature was implemented by Compellent years ago with the name of Data Progression and it is one of the factors of success for this company. Automated tiered storage is a feature that allows to move data blocks between different types of disks and raid levels, in function of their frequency of access, in a transparent manner to the hots. The big advantage is to gain tier 1 disk space and performance moving unused data to tier 2 o 3! With this feature SSD’s really make sense in the real world!

Data Progression from Compellent promises (and honours) big saves in TCO because the automatic tuning of LUNs, ease of use, and less hardware/power needs. Customers are indeed asking for these kind of features every day to lower TCO and improve usability of their systems!

what is doing EMC about automted tiered storage? what does it mean for your TCO? i would like to share you some facts (read from the articles below and no FUD):

  1. FAST V1 (end of the year ?) will work at LUN level, FAST V2 (mid 2010?) will work at subLUN level!
  2. FAST V1 will be incompatible with Virtual Provisioning.
  3. FAST is for V-MAX…. DMX not contemplated!
  4. CX will be supported in the future… CX 4 or CX 5?
  5. SAS backend in the future (means, at least, new CX are you ready ?)

Well, a lot of guesses in my mind 😀

FAST V1 will work at LUN level.. FAST V2 will work at subLUN level.

Yes, you have understood very well! LUN Level means FAST needs to move a whole LUN from one Tier to another every time needed. Others attempted this, years ago, with “automatic volume migration”: the best practices, now, is no more automatic shuffling but to suggest the movements to avoid the performance impact on the Raid groups involved!

Moving TBs of data back and forth can be very dangerous and ends up it limiting usability of that feature.

Next version of FAST (V2) will work in the right way, with finer granularity, and it will move small pages of data so it will be suitable for most needs… BTW, i would like to know if subLUNs migrations will be done only among different types of disks with the same RAID level, or it will be done across disks and RAIDs to achieve the best performance and space usage (without user intervention, of course). Another more thing, how can FAST is reconfigured, on the fly, to rearrange the behavior of the single LUNs? (Compellent does these things automatically when setting the right profile with just a point and click!).

Ease of use and predictable results are key values for TCO!

FAST1, will be incompatible with Virtual provisioning.

It’s not a news… every time EMC introduces new features or functionalities you have to ask yourself about limitations come into play too. EMC, give away Virtual Provisioning for free so it’s up to You to decide if you want a space savvy system or an automated tiered system, not both of them at the same time. what a dilemma!!!!

Moreover, what about snapshots, clones, replicas? i would like to know if there are other limitations about these features? only EMC can answer!

what about a migration impact between V1 to V2? fat to thin+FAST? is it all automatic? V2 (with its 768K pages) new questions arise: where can i put my snapshots? how snapshots impact on performance adding FAST? will snapshot be relocated on less valued tiers to save space and performance?

Every constraint is a limit, each limit force more human work, more work is higher TCO!

FAST is for V-MAX.. DMX not contemplated! + CX will be supported in the future… CX-4 or CX-5? + SAS Backend

Not a problem, DMX customers can migrate to V-MAX only for a “small” amount of bucks, 😉 Probably the same tale will be for actual CX-3 and CX-4 customers. will FAST supports CX4? or you need to migrate to the next CX-5 (or “V-MIN”)?

For other vendors (like Compellent) SAS is an option but what does it mean for EMC customers? Hey, Go and see the rear back of your CX and check if you have any free slot to add SAS backend cards. You are out of luck! You need to buy the next CX generation to have SAS!

Buy new systems means forced migrations, migrations means higher TCO!

So, to recap: Compellent 3 – EMC 0 about TCO in automated tiered storage!!!

EMC adds FAST but V1 has a lot of limitations and V2 will see the light in a year (likely with less limitations). On the other hand, Compellent implements this feature (called Data Progression) since years and it is well implemented, without limitations and constraints!

EMC wants you to buy new hardware (V-MAX and probably next CX) to implement FAST!!!! With Compellent you may add features to every generation of the Storage Center without chancing hardware.

EMC plans a SAS backend in the future… Compellent can mix and match SATA, SAS, FC, SSD backends in the same storage system just now, without costly “rip-and-replace” upgrades.

Hey, ok, i’m a Compellent fan and reseller so i’m not really unbiased, but every comment is very welcome!!! 🙂


Ti piace questo articolo?

Share on Linkdin
Share on Twitter
Share on Facebook

Lascia un commento

Privacy Settings
We use cookies to enhance your experience while using our website. If you are using our Services via a browser you can restrict, block or remove cookies through your web browser settings. We also use content and scripts from third parties that may use tracking technologies. You can selectively provide your consent below to allow such third party embeds. For complete information about the cookies we use, data we collect and how we process them, please check our Privacy Policy