December 22, 2008

Unified Host Utilities Kit 5.0 for ESX Server is Released

Last week we released a Unified version, 5.0, of the Host Utilities Kit for ESX server. There are some changes, as well as some new additions to the Kit.

First of all, a new script has been added called config_nfs which sets the appropriate NFS Heartbeat parameters.

The filer_info script has been renamed to controller_info and it now supports SSL. In fact, SSL is also supported for the config_hba and config_mpath scripts.

There's a new utility that has been added, called mbrscan. The purpose of mbrscan is to identify wheather or not a VM has properly aligned partitions. The script will run successfully under two conditions: a) The VM is powered off b) the VM has VMsnaps. The script will error out if none of the above two conditions are satisfied simply because the VMDK(s) are locked.

There are new utilities added that will you help set GuestOS timeouts. That includes Windows, RHEL, SUSE and Solaris.

Here's a list of the all the scripts included as part of the Unified Host Utilities Kit v5.0:

  • install - Install script for the EHU
  • brocade_info - Collects configuration information about Brocade FC switches
  • cisco_info - Collects configuration information about Cisco FC switches
  • config_hba - Utility used to set HBA parameters for communicating with NetApp storage devices. This script will get run as part of the installation and can be executed subsequently at any time.  Support was added for 8GB FC HBAs and well as FCoE CNAs.
  • config_mpath - Utility used to determine which of the available paths are primary paths and to set primary paths
  • config_nfs - Utility used to set the NetApp recommended NFS Heartbeat settings.
  • controller_info - Collects configuration information about NetApp storage devices.
  • mbrscan - Utilty used to check vmdk files for proper alignment from the ESX console (for VMFS and NFS datastores), and from unix/linux (NFS datatstores).
  • mcdata_info - Collects configuration information about McData FC switches
  • qlogic_info - Collects configuration information about QLogic FC switches
  • san_version - Prints the EHU version
  • sanlun - collects information about the LUNs currently mapped to your host
  • uninstall - Uninstall script for the EHU
  • That said, I'd like to wish Happy Holidays to everyone.

    December 12, 2008

    Shooting Fish in a Barrel

    According to a figure posted in September by Robin Harris over at StorageMojo, NetApp has 13,000 systems running deduplication. That's an impressive number especially for a feature that was officially announced in May 2007. In fact, the NetApp deduplication license is the most commonly generated/downloaded license in the company's history. The fact that it is free, probably has played a role, but more importantly, it also points to the fact that customers are  looking to deduplication to address increased storage costs by adopting storage efficiency technologies.

    The latest, up to date numbers, I just received are depicted below.

    Pic3_3   

    Many of you that pay attention to this blog know that in September 2008, we announced a 50% Guarantee program leveraging some of these storage efficiency technologies. I'm not going to go over the details of the program nor am I going to debate the points the competition has raised. This has already been done here, here and here .

    Instead what I'd like to talk about is how did the Guarantee come about plus I'd like to provide some data.

    Similarly to other other storage vendors, we collect array information by an array "call home"  capability. We call this AutoSupport (ASUP). Using AutoSupport we can examine various things about the array, such as its status, configuration, licenses, and general performance information.

    A while back, engineering noticed that our install base was achieving very high space savings because of deduplication. In fact, the numbers were as high as 80-90%. Needless to say, this knowledge got circulated around the company and as a result of that Marketing, with the support of our Legal department, thought it would be beneficial to our customers to get the message out and put together some a Guarantee program. So the 50% Guarantee Program was born. By all accounts, this is a conservative number but it is a number that makes Legal happy although we know that the space savings achieved are much more than that.

    In fact, we have the customers to prove it. Customers such as Burts Bees. You probably have no idea who Burts Bees is, but I bet your wife probably does. Mine did. There's a great article by Ted Hein on Burts Bees initiative for Green Storage titled "Burts Bees: What IT Means When You Go Green". I highly recommend reading the entire article. There are some very interesting comments in the last section titled "Doing the Math". Like this one:

    "Combined with NetApp flexible volumes, data deduplication has allowed us to roughly triple the effective capacity of the SAN. Beyond VMware, by moving our physical servers’ storage onto the NetApp platform, we’ve been able to reclaim roughly 300GB to 400GB for every terabyte of storage found on the original server."

    The hardware and virtualization savings alone paid for our green IT initiative. The rest came with the solution: energy savings, improved fault tolerance, recoverability and performance. "

    Get outta here!!!! "Roughly Tripple"? I don't believe it!!! Chuck said it's all BS, it's trickery, and so do the HP guys who gave us the "Real 50% Guarantee Story".

    How about this article on the  Duke Institute for Genome Sciences:

    "NetApp deduplication also played a key role in harnessing data growth at IGSP.  NetApp has saved us a lot on storage for VMware. When we originally set up VMware, I allocated about 2.4TB for it. With NetApp deduplication, I’ve been able to shorten that down to less than 700GB. We now see an average of 83% reduction in redundant data on our VMware systems."

    After I read the above articles, I then started searching around the net for users posting their experiences with NetApp deduplication and i run across StorageMojo's entry on deduplicating primary storage. There are some interesting comments there. In fact, one of the users has posted output from this array:

    Pic1

    Then as I moved on, another user posted the following:

    Picture3

    People get hang up on the 50% number and forget that this number is, by most accounts, a conservative estimate. Potential users need to keep in mind the larger picture which is, thousands of NetApp customers are running deduplication today achieving substantial space savings.

    Some of our competitors have criticized us because of this program. In fact, 2 months later some of them keep posting blogs about it (HP). That's a good indication of a “deduplication heartburn”... Like they say, where there's smoke there's fire. Others, we're told, have issued memorandums to their sales force utilizing clever U-Turn techniques in order to shift the conversation into other areas where they can compete.

    We've opted to use customer references instead, rather than waste our time responding, simply because people appreciate other users' experiences more than they appreciate a vendor's hearsay .  We let our customers do the talking for us. It's that simple...like shooting fish in a barrel.

    December 08, 2008

    NetApp Technology Network Solution Contest

    I apologize because I haven't had a blog in a while because I was overseas for vacation.

    I'm a huge soccer fan so I went to watch my favorite Greek soccer team (Panathinaikos)  play in Milan (Inter) for the Champions League. I had a great time, with great seats and what made it even more special was that we won. Img00031_2

    I'm back now and after parsing thru 480 emails I can now post a small blog.

    For those of you who are not aware, NetApp has established, for some time now, User and Partner Communities. (aka NetApp Technology Network).

    There's a lot of activity there among users, partners, NetApp SEs and internal NetApp developers. If you have a question for which you don't know the answer to, the community is your friend as there's always someone who has the answer be it another user, a partner or a developer.

    The NetApp Technology Network has recently announced a contest to recognize users or partners for solutions or tools  they've developed to solve day-to-day operational issues.

    All of the submissions will be made available to all community members and the winners selected will have a choice of cash, for some of the top tools, or some electronic gadget, namely iPODs.

    I would encourage everyone to participate. The goal of the contest is to promote collaboration among community members and help peers solve issues by leveraging tools already in existence.

    You can find more information about the contest here.

    November 18, 2008

    Note: Using esxcfg-module to set ESX Driver Timeouts and Queue Depth

    It is a common occurrence for people to change the HBA driver timeouts and in some cases, the Queue depth settings on ESX server to values suggested by their Storage vendors. On ESX server this is done by using the "esxcfg-module" command followed by a subsequent server reboot for the new value to take effect.

    One thing though that people don't realize is that the parameters are not additive to what already has been set.

    For example, if I had previously set the Qlogic driver timeout parameter, qlport_down_retry, and then at a later time I decide to change the default queue depth parameter, ql2xmaxqdepth, I must specify both parameters as part of "esxcfg-module" command otherwise the qlport_down_retry value will revert to its default value.

    So the proper way to set either one of the values, even if you need to modify one of them is to specify both. This applies to both Qlogic and Emulex.

    #esxcfg-module –s "qlport_down_retry=X ql2xmaxqdepth=X" qla2300_707_vmw.o

    For NetApp, the HBA driver timeouts are set by the ESX Host Utility Kit (HUK) and depending on ONTAP release, the driver timeouts range between 10" and 60". Anything above ONTAP 7.2.4 (inclusive) using Single System Image Cluster Failover Mode uses a 10" timeout. This setting applies to both Qlogic and Emulex.

    November 11, 2008

    Windows 2008 Hyper-V and SCVMM 2008 Part III

    I didn't have a chance to continue writing last week on my Hyper-V series post because I had to deal with Domain issues that prevented me from adding my second Hyper-V node to the domain. Finally, I was able to figure out the root cause which was duplicate SPNs and fix it last night.

    In the last two posts I talked about some of the Hyper-V requirements, and dealing with Server Core installations. In this post I'd like to talk a bit about SCVMM (System Center Virtual Machine Manager).

    Continue reading "Windows 2008 Hyper-V and SCVMM 2008 Part III" »

    October 30, 2008

    NetApp & 10GbE support - VMware Support Matrix

    Although, Fibre Channel continues to dominate with respect to ESX server deployments, IP is seen some very high growth rates in this space.

    Over the past several months we've gotten several requests for 10GbE support. The challenge was that although we have had 10GbE capability within our arrays for some time now, from an ESX perspective, VMware didn't support 10GbE connectivity from the VMkernel to the back-end storage device. The only VMware support that existed was for "front-end" connectivity, from the VM(s) out to the network.

    As most folks are aware, with the introduction of ESX 3.5U1 VMware announced support for 10GbE back-end connectivity (VMkernel to storage array). What does this mean?

    Am I supported?

    The fact that VMware announced support for 10GbE does not mean any array with 10GbE NICs will be supported. The array must be on VMware's Suport Matrix for 10GbE connectivity.

    Rule of Thumb: If the Array uses a 10GbE interface to connect to an ESX server(s), regardless of the negotiated speed, the array needs to be on VMware's Support Matrix.

    Based on yesterday's update of VMware's support matrix ,we're now listed  for 10GbE storage connectivity to VMkernel. Although, we can run same version of ONTAP across the product line, each HW platform needs to be separately qualified.

    As it stands right now, it looks like we're the only vendor on the matrix able to receive VMware support and we're working quickly in expanding support to include other arrays that will benefit from it.

    October 29, 2008

    Windows 2008 and Hyper-V Part II

    In the previous entry I discussed the different OS versions and  requirements for enabling the Hyper-V role. In this post, and given that Microsoft recommends Windows 2008 Server Core for Hyper-V deployment, i'm going to discuss it.

    Continue reading "Windows 2008 and Hyper-V Part II" »

    October 25, 2008

    Windows 2008 and Hyper-V - Part I

    I've spend an enormous time over the years with VMware's robust solution but it's time to dedicate time to other virtualization solutions as well because some of them are very deserving. Hyper-V's an example of that and frankly so is XenSever. That said, in the next few blog entries I'll be talking about Hyper-V, its components and Hyper-V on NetApp.

    Continue reading "Windows 2008 and Hyper-V - Part I" »

    October 14, 2008

    The BT Consolidation Story

    Two years ago, Michael Crader, the head of Windows Consolidation for BT kicked-off a very ambitious project to consolidate more than 3000 Window servers.

    Back in 2006, BT was facing exploding costs thru its power consumption of close of 0.7% of UK's power production. BT also had severe problems by failing to maintaining its SLA given that backups were taking days (96 hours to be exact) and new server deployment could take six weeks among other things.

    Continue reading "The BT Consolidation Story" »

    October 13, 2008

    Is the Future of Fibre Channel, Ethernet (FCoE)?

    A title like this one four years ago would have been joked at. However, today, given the industry buzz, hype, and genuine excitement that FCoE (Fibre Channel over Ethernet) has generated, it seems as a very reasonable assumption and one that will, most likely, come to fruition given the support FCoE has garnered support from pretty much everyone in the industry.

    What is FCoE?   

    Fibre Channel over Ethernet or FCoE, is a new protocol (rather transport) undergoing standardization at the T11 committee. FCoE  transports Fibre Channel frames over Ethernet by encapsulating FC frames. Although the standard is not expected to be ratified until the end of 2008, the expectation is that no spec changes will occur between now and the then.

    Continue reading "Is the Future of Fibre Channel, Ethernet (FCoE)?" »

    © NetApp, Inc.  |  "Safe Harbor" Statement