Summary – this ended up being a good summary of the highlights of the keynote later in the day – see my live blog post for details on that (not just text…lots of picture there too). I always like hearing what executives choose to focus on and how they say it when only have 2-3 minutes to summarize a huge amount of items.
EMC Redefine Possible – Live Blog
This will be a somewhat live blog although posted later (that is, not analysis per se….noting what was stated, asked and answered). Panelists were Bill Richter (Isilon), CJ Desai (Emerging Technologies), Fidelma Russo – VMAX, Stephen Manley – CTO of DPAD
Focus for this session is on today’s announcements and their impact and the business units – not roadmap (that’s for other times). Each exec is first giving an overview.
Bill starting – 4 things
- 2 brand new hardware platforms – S210, X210. This is significant b/c we don’t do too many Isilon hardware releases – 2-3 years since last hardware release so this is a big deal.
- New version of OneFS is huge – ‘flash is cache’ is huge.
- Brand new access methods – SMB multi-channel – big deal for media and entertainment sector – OpenStack Swift direct support.
- Brand new solutions – both within EMC and also across the Federation. VCE, etc. Also a joint solution between XtremIO and Isilon within VCE for VDI – scale-out for both Flash and File for VDI
CJ Desai – introducing Josh Goldstein (runs Product Management and Marketing for XtremIO). After GA in November XtremIO has exceeded all EMC expectations. 2.4 focused on snapshots (innovations in delivered), Data at Rest Encryption, etc. 3.0 will be out in a couple months – HUGE feature is inline compression (Andrew’s note – this is truly huge) – supported with all features – allows addressing many more traditional workloads (databases, etc. is the goal). Moving very aggressively on the roadmap. Also a starter edition of XtremIO at 5 TB (half X-Brick). Cluster expansion up to 6 X-Bricks coming.
Fidelma Russo – VMAX focus. Refreshing the entire VMAX product line – going to 100k, 200k, 400k. Ivy Bridge, fully SAS, Infiniband faster. Entering lower and scaling higher from an engine perspective. Really around new OS – calling it HyperMax. Vastly increasing the amount of metadata that can handle in order to give cloud scaling capabilities. Doing fully dynamic core allocation – can dynamically shift from front-end to back-end based on workloads. Large machine is up to 384 cores – lot of compute power. Newly redesigned snaps in the data servers for TimeFinder – scales up to 16 million snaps. Zero space snaps until restored. Embedded storage hypervisor in the box to embed other applications within VMAX – examples are embedded file based on VNX code (compatibility across transactional file), embedded management such as Unisphere, embedding FAST enhancements around database and pattern recognition. Many more features…more than can cover in 60 seconds. Announcement – acquired cloud access company called TwinStrata. Working closely with Data Domain team – use new snap technology to use Data Domain as target – bypassing backup servers, backup network, etc – this is called ProtectPoint.
CTO of DPAD – talking for years about how ‘backup is broken’. Driving multiple initiatives around that – snapshot-based incremental backup, tighter integration with VMware, etc. ProtectPoint is the next stage in that – solution that lets backup happen without needing any traditional backup architecture. Removes weight on the infrastructure and crushing application/database servers – data moves from VMAX to Data Domain directly. Data stays in native format – not a tar format or otherwise. Data protection happens in so many places in the infrastructure – goal is to give app teams and storage teams control without stepping on each others toes. Goal = data protection becomes a feature embedded in the infrastructure – not a standalone feature.
Commentary on TwinStrata acquisition – everyone is looking at hybrid clouds – how do you build a gateway between on-prem and off-prem? Looked at many companies and saw this one as best – also local to MA which is good for VMAX team as well. Goal = giving customers choice especially around cold data that want to handle with a policy-driven approach to move off-prem in various ways. Planning to use FAST technology to drive this.
- Will Twinstrata stay OS-agnostic/target cloud agnostic? Answer = no plans to change anything.
- What do you see as the Total Addressable Market for this? (rephrasing the question) Answer = all enterprise customers are talking about this. Also looking at this in the Isilon area as well. Around the world everyone is trying to come up with their public cloud or off-premise strategy. Theme of EMC is around choice for the last couple years – this continues that theme.
- Technology transitions – much talk about all-flash datacenter – but also talk about a hybrid strategy – when do you foresee the death of all hard drive arrays? (no flash) Answer = that’s already happened pretty much on enterprise storage. Only place is still hard-drive arrays is on the mainframe side – seeing adoption of flash and array tiering in the mainframe arrays increasing this year though. It’s happening in Isilon soon – DPAD will likely be last just for practical reasons.
- Given increasing focus on flash portfolio, what are you doing to right-size the amount of flash for customer needs? Answer = very good sizing tools already, to some extent don’t know all customer workloads and how customers will use this (i.e. customers can innovate too). From a OneFS perspective, will always get 100% utilization of SSD’s. VMAX3 – tied to SLO provisioning, based on tons of data from VMAX and current FAST VP tiering. DPAD – no one wakes up and asks “how can I balance flash vs. disk today?” – amount of data that VMAX team has on this is a researchers’ dream.
- If I’m the head of IT, how do I sell this to my management board who don’t ‘speak IT’? Answer = VMAX traditional values are power, trust, smart – have added agility. Value to business is more agility – agility in deployment/provisoning, agility in consuming it, etc.
- Followup – can you say that in other terms? Answer = all about lower cost of ownership. Summary = cost-optimization and first step to Business Intelligence (consolidate your data – then with EMC partnerships with many other companies easier to start getting value out of the data).
- Foskett asking – what about VIPR and TwinStrata? Answer = it’s day one…and we’re all here. But…in general, yes – no idea of details yet.
- David Ring asking – as data/workloads grow, will there be any remediation to the SLO? Algorithms within HyperMax – places you in best place possible within SLO. Will fill up IOPs and/or capacity (but not to detriment of higher priority SLO’s). Workload planner in Unisphere so can model out. Also doing FAST Sideways.
That’s a wrap…heading over to Old Billingsgate Market now for the actual event.