EMC Redefine Possible – Live Blog

Note: this is a live blog…apologies for typos – this will be relatively analysis-light (focused on keeping up).

IMG 5099

Walked in to live music and customer quotes on the big screen – one of them was TIAA-CREF (customer who Varrow works with in Charlotte, NC).

3D Glasses on all the seats – opening video uses 3D – quite cool.

David Goulden coming out of the Tardis (blue telephone box) – CEO of EMC Information Infrastructure.DSC 5235 Major challenges are 1) Existing Workload Growth and 2) Infrastructure Application Drag (58 cents of every application dollar are spent on backend infrastructure drag – backups, replication, etc.), 3) Performance Hungry Workloads.

And what about new applications? What are you investing in? New apps have major differences in their infrastructure demands – but we don’t want to create new silos for new apps – we’ve all been there and know how painful that is long-term.

Big announcements today – XtremIO, VMAX3 (brand new family, Dynamic Virtual Matrix, HyperMax OS), VIPR 2.0 and SRM 3.5 shipping and available – free for first 50 TB, Isilon (next gen of Isilon systems and OneFS file system.

DSC 5270

First shipments of ECS are to the Vatican – 3 PB Object Store (one full Rack). 

Jeremy Burton now on stage – ‘No instructions on the inside of the Tardis – almost got stuck pushing instead of pulling on the door.’ First – XtremIO…

DSC 5273

  • In six months went past $100 million of revenue for XremIO – fastest ever revenue growth for EMC (faster than Isilon, than VMware, anything). Why so fast? Because architecture matters – ability to deliver low-latency with features inline. 40% of customer base is more than a single X-Brick. Seeing 99.999% availability from field data.DSC 5305
  • Baby X-Brick – 5 TB – non-disruptive expansion from 5 TB to 10 TB.
  • Now going to 6 X-Bricks as well.
  • Data services – always inline, all the time. Before this was Thin Provisioning, Deduplication, Data Protection. Today we add At Rest Encryption (Hardware Accelerated, Existing or New Hardware if have SEDs. Boom! Adding Data Compression – seeing 2-4x results. Finally – writable snapshots. And now doing a live magic act – swallowing live needles because that’s related to snapshots. Writable snapshots previously had issues with metadata bloat. XtremIO = instant creation, instant everything (I missed the whole list).DSC 5309DSC 5344
  • Walking through a Prod/Dev/Test scenario – a 10 TB prod database can easely balloon to 100-200 TB of actual data on storage arrays. XtremIO Example – 19 DB copies of a 10 TB database in 19 TB.
  • Miroslav Klanisky demonstrating writable snapshots with Jeremy. Cloned many Prod DB VM’s – using SLOB to generate workload. One VM generating 60,000 IOPS of traffic under a millisecond.DSC 5367
  • All the new data services are covered by the same $1 Million guarantee they will always be live/inline.
  • Major tweaks to current code to improve performance – 2x faster SSD rebuilds, 50% faster DB performance, a couple more.
  • Integration points – VPLEX for RecoverPoint, ViPR for object and other.
  • Interview with Boston Scientific – DSC 5399
  • XPect More Program – 7 year maintenance protection, 3 year money back guarantee, 7 year flash endurance guarantee.DSC 5410
Now on to VMAX3…
  • Models change to 100K, 200K, 400K
  • Dynamic Virtual Matrix – up to 384 cores on a fully built out VMAX. Many things require CPU and memory inside the VMAX – can switch cores between front-end ports and backend ports.DSC 5420
  • HyperMax OS can run not just data services but also infrastructure services – storage hypervisor on the array itself.
  • Running things on the storage hypervisor – data-centric workload services. File gateways, cloud gateways, FAST algorithms. This can broaden out to things like VPLEX, replication, ETL tools – much more coming over time.
  • HyperMax OS = ‘a little bit of magic’….good excuse. Magic trick of a phone magically appearing inside of a balloon….fun stuff.DSC 5424
  • Most importantly, a cool new VMAX rack.
  • Cool higher maximums – New VMAX maximums – 16 TB Cache, 384 Cores, 5760 drives, 6.4M IOPs, 1792 Gb/S
  • VMAX vs. HDS – 2x more drives, 3X more cores, 3x more density, 6x more flash, 8x more cache.
  • Expressing it in Virtual Machine terms – think that a VMAX 400k can go up to 40,000 Virtual Machines (5,000 VM’s for VMAX 100k, 20,000 VM’s for VMAX 200k).
  • Base of 6 9’s with VMAX, 7 9’s with VPLEX
  • Traditional Backup = workload hated by datacenter admins. Very true.
  • Introducing ProtectPoint – why not let the primary storage array drive the copy of the data.DSC 5439
  • Now into a demo of Oracle Enterprise Manager with ProtectPoint.DSC 5444
  • Agile – not a word usually associated with Symmetrix (more about powerful, trusted, stable, etc.).
  • New Service Levels (SLO’s) as part of VMAX – Diamond, Gold, Silver, Balanced. Vince from the VMAX team now demonstrating this
  • SLO Modeling tool inside Unisphere – can check ahead of time if can guarantee service level. I saw a deeper dive on this previously as part of the VMAX beta program.
  • See which workloads are ‘Stable – Marginal – Critical’ in meeting their SLO guarantees.
  • Now a customer interview with USAA – Fidela Maruso introducing Rick Webster with USAA (IT Technical Manager, Enterprise Storage)DSC 5451
Now a focus on VIPR and ECS (backfilling pictures during this section). Design point for ECS is public cloud – that’s what it’s meant to be competitive with – not enterprise storage.

DSC 5459


Now Scale-Out NAS…4 main items below after general notes.
  • New Isilon hardware models – S-Series, X-Series – tier of flash to accelerate delivery.
  • Insane amounts of unstructured data growth – 2013 = 37 EB, 2015 = 71 EB, 2017 = 133 EBDSC 5462
  • Bringing the magician back on stage for the final trick – got 3 random bits of data from the audience..then they were written on a piece of paper inside a locked box that had been hanging in front of the audience for the last 1.5 hours.DSC 5476DSC 5485
  1. New Platforms – S210, 2X more IOPs, Ivy Bridge Based. X410 – high throughput, 70% more throughput over X400. 144 node cluster max = 200 Gb/s – applicable to big bandwidth systems (Hadoop, etc.). 
  2. SmartFlash capability – previously could add up 37 TB of DRAM, now can add up to 1 PB of Flash as Cache. DSC 5496
  3. New Access Methods – SMB Multichannel allowing 4k content (Microsoft), HFS 2.3, native Openstack Swift interface coming soon.DSC 5499
  4. Solutions – VDI vBlock including both Isilon and XtremIO – scale-out higher performance block and scale-out NAS for user files.DSC 5503
    1. Hadoop Market Leadership – see this as a big deal. #1 Market Leader in Hadoop Shared Storage, 400+ Customers, 250% YoY Growth. Not just with Pivotal – also with Cloudera and Horton.

And that’s the wrap on everything David Goulden promised at the beginning…that’s a lot of stuff in the last hour.

DSC 5507

Thanks much to Sean Thulin for all the pictures – much appreciated.

2 thoughts on “EMC Redefine Possible – Live Blog

  1. Pingback: Think Meta » EMC Redefine Possible – Analyst Breakfast & Executive Q&A (Good event summary)

  2. Pingback: A conference in Austin? (or I’ll be at Dell EMC World) | Think Meta

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s