Live from VMworld

This week I will be at VMworld and will be giving ongoing reports of what we see there.

Overall,the show seems as busy as last year. The facility is a little smallerand attendance may be a little lighter but they are expecting between10,000 and 12,000 people. Look for quick updates, in reverse order,throughout the week. Updates will also appear at the top of the page.

Akorri provides softwarethat from a storage perspective supplies intelligence to help ITprofessionals the interrelationships between VMware and CPU, memory,networks and storage. The agentless software helps their usersunderstand what virtual machines are using which parts of storage orwhat the interrelationships are between virtual machines and thoseresources that they share. The package helps answer questions where VMloads are coming from and what guests are causing problems in theenvironment. Using this software allows increased VM densities byunderstanding how VMs are contending for resources across CPU, memoryand storage.>

The latest release announced at the showexpands the view ESX cluster wide. They added analytics around thecluster to understand efficiency as well as limits and allows you toanswer these questions; Are you getting what I expected to get? Howoften am I running above the limits of the cluster or is there are wayI can bring in more resources into the cluster? How many VMs can I packon each ESX host without impacting performance? What is the optimalsize of each VM for each application?

In short, this allowscustomers to take those hard to virtualize servers that are log jammingthe virtualization rollout and move to the next stage ofvirtualization. This is done by providing analytics that allow users tounderstand what the performance needs of these critical servers are andto show that the performance deliverable after virtualization has beenmaintained.FalconStor

was talking about their VMware initiatives where they are providingcontinuous availability for VMware. They are announcing at the show afail back plugin to Site Recovery Manager (SRM). It discovers thereplicated VMs, reverses the direction of the replication, reconfiguresthe VM on the primary site and then initiates the fail back. Now SRMfail back is completely automatic.

Second, they announced theability to extend SRM to physical environments. FalconStor will copyphysical workloads, convert them to virtual workloads and mount them asstand-by VMs. From there SRM can take over. This demand is coming fromcustomers who have hybrid environments; some physical, some virtual.

Third,they announced that they are providing more manageability to VirtualServer Admins. They announced a FalconStor plugin to Vcenter. Thisplugin will allow the user from a single interface to create thevirtual machine, assign storage and even assign a data protectionpolicy that includes snapshots and replication.

Finally, theyare extending their storage and data services to the VMwareinfrastructure via virtual appliances. All of the products are nowavailable as Virtual Appliances; the latest is FalconStor'sdeduplication appliance which can now run as a virtual machine.

EMC wastalking about the VMware specific capabilities of its PowerPath VirtualEdition. This brings the multi-pathing capabilities of PowerPath to thevirtual environment. Configuring storage IO paths in VMware can be achallenge to say the least, Powerpath VE resolves many of those issuesautomatically. Simply point the virtual machine at PowerPath VE and itcan automatically manage all the IO pathing for you. This includeskeeping the path configuration in tact as you move VMotion, the virtualmachine. It will self optimize depending on the target and can includeiSCSI in its path management. In addition to managing the failoverpathing it can also manage performance pathing. Instead of the normalround robin approach it can redirect traffic to a specific least busyIO path.Isilonis increasing its presence in the VMware space taking their scale-outNAS capabilities to the server virtualization market. As we talkedabout in our article "What's Causing the StorageIO Bottleneck", VMware is one of those environments that hasa lot of storage IO requesters (Virtual Machines and Physical Hosts)generating highly random workloads. Isilon can deliver all the benefitsof NFS mounted VMware virtual machines and by leveraging the scale-outnature of their environment, make the storage environment as agile asthe server virtualization environment.

Storspeed providesapplication aware caching. Application aware caching allows customersto separate the purchase of performance from the purchase of capacity.This can be done by leveraging and extending caching technologiesacross multiple tiers of storage. The Storspeed SP5000Application-Aware Caching solution installs between clients and thefile based storage systems. It uses advanced packet inspectionprocesses to analyze all of the traffic flowing to and from thestorage. With this knowledge it takes the most active blocks of storageand moves them into a three node RAM based Tier 0 storage cluster.

Applicationaware caching makes sure that your most active data is always comingfrom the fastest possible storage tier. What makes this solution reallymake sense is that it is dynamic. It is constantly optimizing througheither its own analysis or your policies to make sure that the mostactive data is always being promoted up the storage tier.

Asigra is an enabler ofservice providers to offer cloud backup services who charge theircustomers a monthly subscription. These providers have the ability toprovide the service in a private fashion as well, so end userorganizations can have the operation on-site. They can then choose tomanage the products themselves or allow the provider to do it for them.At the heart of it is an agent-less backup application thatdeduplicates, compresses backup data and then replicates that data to aDR location or the provider.

New for this show is that theback-end repository is now freed from having to run on a separatededicated server, it can now run as a virtual appliance within theVMware infrastructure. Mobility is a key function of cloud computingand virtualization so it makes sense that the Asigra repository is inline with that model.Data Domain is talkingabout their ability to boot VMs from the DR replica system. As most ofyou know Data Domain stores and replicates deduplicated data, sobackups can be retained onsite and efficiently moved to a DR location.Using a utility or script that protects the virtual machine by copyingit in its native form, you can move the VMDK and configuration files tothe data domain system via NFS. Then in the DR site the VMDK files canbe mounted and VMs booted from an NFS data store configured on thereplica Data Domain system. This is an ideal way to bring servers backonline for file recovery from within the guest OS, DR testing or in thefirst minutes of a disaster recovery plan. If the time at the DR siteis going to be extended then the virtual image can be moved to a moretraditional storage using storage VMotion.

Final briefing of theday is 3PAR.They are delivering storage solutions for serious VMware environments.Most of the ease of use storage solutions that are at the show arefocused on using NFS or iSCSI, both viable but also viable is if youcould make fibre channel easy. 3PAR does this with an autonomic storageplatform that goes beyond the thin provisioning that we discuss in ourlatest whitepaper but adds self optimization, self tuning, is a scalablearray that grows as opposed to an array without adding separate boxesthat increase management complexity.

Combine this with thestandard capabilities of fibre channel SANs like NPIV and the basic QoSfunctions as we discuss in our article "UsingNPIV to Optimize Server Virtualization's Storage" and fibrechannel especially in the enterprise is still something to consider.

CORAID - SMB Storage forVMware, one more time, now with an ATA-over-Ethernet (AoE) twist. AoEis a block based protocol layered directly on top of ethernet. ATA diskcommands (ie. read disk sector x, write disk sector y) are put directlyinto standard ethernet frames using the AoE protocol. Also is anon-routed protocol, therefore does not require IP or TCP protocollayers. This eliminates unnecessary processing and makes networkconnection to disks simple. For ESX CORAID has developed a specializedNIC card that identifies itself as a SCSI card, otherwise ESX wouldassume control of it like it does with any other network card. The cardcomes in two flavors, dual 1GB port and dual 10GB port. The ports areautomatically trunked together providing either 2GB or 10GB ofbandwidth and with no IP overhead performance should be sustainable.

StorMagic - SMBs shouldbe feeling the love. StorMagic is also vying for the SMB market.StorMagic wants to make the storage internal to your servers sharable.With the StorMagic solution each server can see the other server'sstorage as if it were on iSCSI network. The software also has theability to synchronously replicate the data to one of the otherservers, so functionality like Vmotion will work. The use case would bea dual or three physical server environment, where each server wouldreplicate one of the others providing full redundancy and full use ofVMware's advanced features. SMBs have some series cost saving, storagesimplifying choices available to them, no excuse not to deploy VMwarein the small business now.Virtensys is an I/OVirtualization solution. Think of your typical server rack; each serverhas multiple network cards and multiple storage HBAs. In addition theymight have specialized cards that do deduplication or compression. Theproblem is that not all of these servers can justify the bandwidth madeavailable by high end cards. With Virtensys you could put a PCIe card(or two for redundancy) in each physical server and run one cable tothe Virtensys top of rack switch which could contain multiple 10GBEthernet, FCoE, 8GB fibre or any other PCIe adapter. The switch thenallows the I/O to be shared across those 16 machines. Further it letsyou present multiple virtual NICS or HBAs to each physical server. Thenin the VMware use case you can assign these virtual NICS to specificvirtual machines.

The payoff is a significant reduction incabling costs and even more importantly a significant reduction in NICand HBA costs. This is especially true when you factor in the need forredundant cards. In a 8 server rack you would have 8 secondary HBAs and8 secondary NICs. Compare that to the cost of one extra of each typethat effectively acts as a global spare.Pivot 3 is reversing thetrend that I see in some companies who are putting storage applianceson VMware. What if you went the opposite direction and started usingthe excess compute resources on the storage system to host virtualmachines? That would reduce data center footprint!

Pivot 3 is agrid based storage system using multiple Intel based nodes withinternal storage that are clustered together to present a singlestorage array to attached servers, except those servers can be on thenodes as well. The ideal market is storage capacity heavy VMs asopposed to compute heavy VMs. There is a lot of these workloads; backupservers, Exchange, Sharpoint or even for DR. When you need to supportcompute heavy VMs those can be put on an external physical server butstill leverage the cluster for storage as well as an emergency Vmotiontarget.

Marathon

offers high availability simplified in a Citrix XEN Environment, idealfor the small business. It allows the SMB to implement cost effectiveserver virtualization via XEN and then with Marathon's everRun,interconnect another physical server via a dedicated link, for completevirtualized server redundancy.

You can do this in differentsites but it has to be close by since the communication issynchronous.Commvault Using theconference to emphasize their virtualization support. A lot of the hypeof Simpona 8 was on deduplication; virtualization was missed in thediscussion. Off host backup was through VCB. Nothing new there, butwhat is unique is the block level incremental capabilities, so only thechanged blocks within the virtual machine need to be backed up. Theyalso have the ability to auto-discover newly added VM's throughout thelifecycle of virtual machines. When a new virtual machine is introducedit is automatically added to a default virtual machine backup job.Restore,they have a lot of portability in the restore. They have a single steprestore, they support a backup to disk, migrate to tape and thenrecover directly from tape when the recovery need arises. Can restorefrom virtual to virtual environment, for example from VMware to Hype-V.Also they can restore from virtual to physical or physical to virtual.Finally they have tuning parameters that allow fine grained controlover how much performance they will extract from the virtualinfrastructure.

APTARE is a private, profitablecompany. They are primarily a portal for managing and reporting on yourstorage in the enterprise. Over the past few quarters they havereleased modules that include backup manager, capacity manager,virtualization manager and replication manager. They announced 7.1 atthe show and it will be available on September 25th. The tools focus onVMware handling virtualization specific problems like lack of planning,limited enterprise view and VM sprawl.

While all of the moduleswere upgraded, specific enhancements in APTARE's Virtualization Managerfocused on reporting on VM Density, detailed reporting on performancemetrics from VMware vCenter including IO/sec and latency, VM tophysical LUN mapping and reporting, IO Stats and latency from VMthrough to the LUN.

EvoStor, aVMware only, first that I have seen, storage solution. This is morethan focus, this is exclusive support of a single environment. Thedownside, today you couldn't use it for stand-alone serverapplications. The upside is that for VMware environments it haspossibly the tightest integration of storage into that infrastructurethat I've seen. The system is basically a storage grid similar to otherscale-out models like LeftHand or Isilon but again, focused on VMware.

Vizioncore,where we spoke about the new version of vRanger Pro for VMware backups.It can do incremental backups of virtual machines, reducing backupwindows and lowering the costs of backup infrastructure. It also cannow do direct to target backups, which further reduces resources. In atime where customers are looking to reduce costs, vRanger Pro canreduce much of the backup infrastructure investment.Inaddition, they were featuring their partnership with iLand. Bycombining Vizioncore's vReplicator product, iLand's data centersprovide a turnkey, off-site DR solution. This is an ideal combinationfor small to medium size companies that are looking for DR capabilitiesbut don't have a suitable remote site. This allows them to off-sitetheir virtual infrastructure without making the off-site investment.

Storwize, yesStorwize was there. In-line compression works very well on NFS mountedVMware images. They showed it live in the Isilon both, 96 VMs, 70% datareduction, better performance. They are doing for storage for whatVMware did for servers. Which, by the way, I am speaking at a lunchtime on Tuesday and Wednesday.  The obvious payoff is toreduce storage need in VMware environments. The less obvious, possibly,is decreased disk and tape backup space needs, reduced replicationbandwidth and reduced snapshot reserve space area.