Start a Conversation

Unsolved

This post is more than 5 years old

16924

August 15th, 2013 04:00

Ask the Expert: Block Deduplication and FAST Suite Tiering in the New VNX

YOU MAY ALSO BE INTERESTED ON THESE ATE EVENTS...

Ask the Expert: Multicore Optimization (MCx) in the New VNX

Ask the Expert: Data Protection for Critical Applications on Next-Generation VNX

https://community.emc.com/thread/180183

By combining the new  block deduplication in the next-generation VNX  with FAST Suite tiering technologies, you can dramatically lower the costs of the flash tier and drive down the $/GB. We call this Compounded Efficiencies.

In this Ask the Expert event we'll answer your questions on these technologies:

  • FAST Suite improvements
    • The next-generation VNX series increases FAST Cache capacities, delivers 4X better FAST VP granularity, and supports new FAST VP eMLC SSDs to lower the cost per gigabyte.
  • Fixed block deduplication
    • This new VNX feature has a granularity of 8K block size, providing customers with space savings at the production level when data is either redundant, such as extra duplicate copies of the same virtual machine operating systems, or maturing from a constantly used state to a near stagnant state.

Your hosts:

profile-image-display.jspa?imageID=8406&size=350 Susan Sharpe is a Consultant Product Manager in EMC's Unified Storage Division.
profile-image-display.jspa?imageID=8387&size=350 Ryan Poulin joined EMC in 2006 in the CLARiiON Support group. He provided assistance on the CX,CX3,CX4, and VNX series of arrays with a primary focus on  performance. In December 2012 he moved to the Corporate Systems Engineering team, where he has been mainly focusing on the Next-Generation VNX release.

This discussion will be moderated through September 16. Follow the thread in your inbox (see option on right) to receive updates!

In support of this discussion we include these latest White Papers:


467 Posts

September 11th, 2013 21:00

dynamox wrote:

very few changes to Navi Analyzer, that tool has not changed  in more than a decade. Look at performance tools that have been developed for VMAX (SPA and now Unisphere for VMAX). So many metrics, very nice export options (UPV/BTP). Analyzer is so 1990's (thank god you include it now in base package). VNX Monitoring and Reporting has a long way to go.

Why so little focus on developing a more robust performance analysis tool ?

Yes,  the vnx lacks pretty bad in the performance monitoring side of things...

261 Posts

September 12th, 2013 06:00

ayas wrote:

Hi there ..

We hear a lot about BLOCK deduplication in NG VNX , but what about File level ? I would like to know about File deduplication as well. Can I apply deduplication against LUNs assign to DMs ?

If it is not good .. or not supported ... When will this be supported ?

Other vendor products are tend to have Block level deduplication , It would be nice to cover NAS area by EMC !

thanks !

Block Deduplication is not reccomended for use with File Storage. As the LUNs become Thin LUNs when Block Deduplication is enabled, this recommendation goes hand in hand with the recommendation of not using Thin LUNs with File.

The File side has an existing Deduplication and Compression feature that has been around for a number of releases. Single Instance Storage is the Deduplication technique on the File side. This means if the contents of a file are exactly the same, the file is single instanced and all references of that file will point to the one copy the software keeps on the system.

261 Posts

September 12th, 2013 06:00

阿超_SteveZhou wrote:

sorry, i don't really understand the word "spreads out" here, what would the original FAST VP statistics gathering & analysis processes work? Each pool has its own set of processes, or just threads?

Each Pool does its own statistics gathering and analysis. This data is then used during the relocation window to decide which slices need to be moved.

By "spreads out" I mean that the processes for each pool run at different times. They will be spaced out as to not run all at once.

261 Posts

September 12th, 2013 09:00

SjurAndre wrote:

Hey RyanP

RyanP wrote:

Multicore Cache and Multicore FAST Cache are Block Deduplication aware, which means only single copies of data from a pool can end up in Multicore Cache or Multicore FAST Cache.

If 5 LUNs all had the same 8 KB block in Multicore Cache and Multicore FAST Cache, when these 8KB blocks are deduplicated, only a single instance of each will be left in each cache. With this said, Multicore Cache still promotes data in 64 KB blocks.

Will there be any background processes that handles the lefoverspace when the blocks get deduped? Like any kind of reorganisation of the blocks? So they dont get fragmented.

The "freed" 8 KB areas due to deduplication will be reused by new 8 KB data coming into the array. This will help fill in gaps left by deduplication. As I mentioned in another post, when the Thin software recognizes a slice is minimally being used and there is enough space on another slice for the data, the clean-up process is called and the slice contents will be moved to the other slice, and the 256 MB slice will be freed.

261 Posts

September 12th, 2013 09:00

eerov wrote:

Are the slices used for tiering static or is there any mechanism to gather the hottest of hot data to fill x amount of slices and leave rest of slices filled with colder data (tiering wise)?

The VNX software, in regards to FAST VP, only tracks the temperature of 256 MB slices and not the data contained within the slices.

Moderator

 • 

6.5K Posts

September 12th, 2013 21:00

hi RyanP

thanks for the update ,,

Now I understand why NG VNX call deduplication ``Block `` it is now clear !

thanks

Aya

60 Posts

September 16th, 2013 02:00

Hi,

The VNX MCx is post processing de-duplication and not in-line. What were the design factors which over-rode one over the other? Was the amount of memory a factor or not being able to isolate an area for dedupe? say like the Symmetrix Dynamic cache partitioning.

I take it the impact to I/O being served to clients is not impacted by the deduplication as it is run on a scheduled basis?

This will come up in conversation from Customers etc so would be nice to have a bit more info on minimising the impact of post-process dedupe or in-line draw backs.

Thanks,

Victor

p.s. Great ATE discussion

77 Posts

September 16th, 2013 13:00

This discussion has now concluded. Thanks to our experts and all who participated!

100 Posts

May 6th, 2015 11:00

Good stuff.

1 Rookie

 • 

5.7K Posts

May 9th, 2015 00:00

there was actually a lecture on this topic during EMC World. I hope you attended!

4.5K Posts

May 11th, 2015 09:00

Tim - did you get to go to Vegas and not tell us?

glen

No Events found!

Top