Ah – I always love this period :-) Finally, I can speak freely about things that have been cooking for a long time (years!) This may sound whiny, but it’s actually hard keep a lid on things this cool, to participate in the alpha and beta cycles, and to be one verbal misstep from getting the NDA police on you is not easy. One gets good at the NDA tapdance….
Today, July 12th, VMware has launched vSphere 5 and the related wave of surrounding releases: Site Recovery Manager 5, vShield 5, and vCloud Director 1.5! Note that the GA date is still a little out – but not much.
While I’m sure there will be broad and deep coverage across the blogosphere, Here’s my 2 cents.
You can also see EMC’s vSphere 5 launch site here: www.emc.com/vmware. It has links to other content you might find useful. Also remember that if you have any vSphere 5 questions that are EMC related – or ANY VMware/EMC questions – you can always go to the “Everything VMware at EMC” community at www.emc.com/vmwarecommunity
We’ve also updated the popular “Top 10: Why EMC for VMware” one-pager – which you can see here.
With vSphere 5 and the release wave there are 3 macro observations in my mind:
- The room for anything to NOT run as a VM has shrunk to zero, or close enough that if you’re wasting time arguing about the fraction of a percent that isn’t a fit, you’re really wasting your own time. Customers have been virtualizing their business-critical apps for some time, but there were always outliers. An example is EMC’s own Oracle 11i apps. With vSphere 4.1, we virtualized the whole app tier. We also moved the database tier from Solaris to a 192-core Oracle 10g RAC cluster running on linux on Cisco UCS. That project improved performance out the wazoo, and saved $5 million dollars (including lowered Oracle licensing costs). And that was with the Oracle RAC DB still on physical hardware. With vSphere 5, we are virtualizing that core Oracle 10g RAC cluster – we’ll be sharing our testing and project results at VMworld.
- The path to cloud is even more clear. Cloud (at least at the IaaS layer) is fundamentally about pooling resources, presenting that infrastructure with a multi-tenancy model, making the infrastructure API programmable, and exposing self-service portals. vSphere 4.x did a good job of making CPU and memory into resource pools, and vCloud director 1.0 did a good job of making network pools and self service portals. The gaps were that: 1) storage was NOT very “pool like” for many customers; and 2) vCloud Director lacked API completeness (as well as other v1.0 gaps). With Storage DRS and Policy-Driven Storage in vSphere 5 and the updates in vCloud Director 1.5 – we have a huge leap forward.
- The virtual world is fundamentally EASIER to secure than the physical one. There are lots of enhancements in vShield 5, but there is no doubt that what people will latch onto (I do!) is the new vShield App 5.0 Data Security feature. This is something we’ve been working on between RSA and VMware for some time (if you dial the “wayback” machine – I demoed a prototype at VMworld 2009 :-) This means that compliance with more than 80 standards (like HIPAA, PCI DSS, SOX, etc…) is as easy as “run it as a VM, and place it in the compliant trust zone”. The fact that that detailed data-at-rest classification, including complex algorithmic pattern matching and full text extraction happens with NO IN GUEST AGENTS is just one example transformative effect virtualization can have on security.
While the first thing people will lock into will be the new core VM limits of the Virtual Machine version 8 (32 vCPUs and 1TB of RAM) and Storage DRS (count on it) - to me, the details are even more compelling, and where I’m going to focus. I’m going to cover this as a series of posts to be able to do it justice…
But – before I do, congratulations to all the employees of VMware!
EMC stands beside you as a technology partner – in full support from an engineering, go-to-market and support standpoint as you folks raise the bar, and change the world, once again!
OK – here are links to my post series on the things that are the standouts to me. Read on!
- vSphere 5 – my highlights (some with their own entire blog posts):
- Single unified ESXi – no more Service Console. Service console is dead, long live the vMA! ‘Nuff said!
- VM HA – a ground up rewrite, and great foundation for the future. This is worth a deeper dive.
- vSphere Storage Appliance – bringing the benefits of shared storage to places it’s never been before – for remote offices, SMB customers and retail use cases. ….And if you like simplicity, VMware integration and low cost of the VSA – but want 10x more availbility, scale, capacity, and want the ability to use the storage for vSphere and any other use case – for just a little more, you’ll LOVE the VNXe!
- Auto Deploy – won’t do a full post on it, but it’s great, and about time :-) Lance Berc – great to see your efforts see the light of day, you and I both know a lot of customers who have been asking for this!
- Massive additions to vStorage and Storage-related functions
- VAAI expansion – 2 new block APIs, 3 new NFS APIs (and what would VAAI be without another new “secret” API :-)
- vStorage Storage API for Storage Awareness (yikes what a mouthful – we just call it VASA)
- VMFS-5
- Storage DRS
- Storage vMotion enhancements
- All Paths Down (APD) gets joined with another scary three letter acronym – Persistent Device Loss (PDL)
- Software FCoE Initiator
- NFS client (while still NFSv3 argh) works right with DNS round robin (handy for scale-out NAS)
- Site Recovery Manager 5 – simplified failback (huzzah!), and more formalized workflows for workload mobility
- vSphere Replication v1.0 – after installing SRM5, this ability to simply and easily replicate at the host layer (which is fundamentally a vSphere service, just exposed via SRM) will bring DR to a whole new market unserved via traditional array-based replication. The ability to dial in a 15-60 minute RPO for a single VM (or groups of VMs all at once) is great…
- …And if you like simplicity, VMware integration and low cost of the vSphere Replication – but want 7x better WAN utilization, and want the ability to leverage the new simple SRM failback, want consistency across more than a single VM, have VMs that aren’t powered on, or templates and other content you need to replicate, or want to avoid copying full VMs for resync and other failback scenarioes – you’ll love how all EMC’s replication technologies – from the smallest to the largest scales are all integrated with Site Recover Manager 5.
- vShield 5 (including vShield App 5.0 Data Security)
- vCloud 1.5
Also – I’ll be maintaining a single blog post on EMC’s core product support for vSphere 5, vShield 5, and vCloud Director, as we get closer to GA.
EMC is of course invested in being first and best with support and integration with every API, every function we possibly can. Of course, while we’re not perfect, and hey, it’s not GA yet, it’s a pretty solid bet that if you as “will EMC support ______” the answer is “YES!”
It’s for those reasons, and of course – our mutual customers’ support – for which we are always thankful, that you can expect the two observations below to continue – and to accelerate.
UPDATED July 27th, 2011. Wikibon did a little more analysis, and got more data from the other vendors, which moved around the number 2 and 3 players a bit. The udpated chart above reflects those most recent corrections.
Great summary Chad and thanks for your continued support. Monumental day for VMware and more importantly a huge day for our customers.
Posted by: Dave Schroeder | July 12, 2011 at 08:34 PM
Thanks Chad.
What about low latency apps, like stock trading platform?
I also think apps that require graphic card is still not a good candidate.
Thanks!
e1
Posted by: iwan 'e1' Rahabok | July 15, 2011 at 10:38 AM
A great achievement no doubt!
P.S. What about MSCS clusters? This is a big area where many people are not virtualizing today due to all the caveats on vSphere from both VMW and MSFT.
Posted by: udubplate | August 03, 2011 at 06:07 PM