United States-English

Around the Storage Block

No, this isn’t your typical Pepsi taste test challenge

Published 03 March 2008, 01:06 PM

-- Buzz Kaas, director of IT, Pattillo Construction

I participated in a HP vs. EMC “taste test” challenge at the HP facility in Houston this past January where we compared the usability of the EVA4400 and the EMC CX3-10.

Working side by side with a SE for each product, we ran through a typical setup of: creating disk groups, selecting RAID levels, creating LUNS, assigning to hosts, several types of virtual cloning routines and tearing it all back out.

Without a doubt, the EVA4400 was easier to use and each task was accomplished in a fraction of the time compared to the EMC CX3-10.

But, more important than ease-of-use is the unique architecture of the EVA. For example, when making a disk group, the controller can auto select the drives for the group. All RAID types (0, 1, 5) are spanned across all spindles in the group. No more hot spots and maximum IOPS for all RAID types. All LUNS are shared across all drives in the group, so all spare disk capacity is pooled together. This makes for a very efficient use of disc space, because you don’t have to have spare capacity designated for each LUN. Expanding a LUN was as simple as typing in the number of GB you want to expand or contract to.

Finally, why risk your data, when you can work on an offline copy in just two minutes? Making clones requires just a few mouse clicks and I can definitely see how we would use cloning for all of our test environments when it’s this easy.

There were three flavors of cloning for the EVA4400:

1) Vsnap: Instantaneous copy using pointers then deltas are accumulated with each page change.

2) SnapClone: Instantaneous copy, but allocating an identical LUN size. SnapClone starts out like a Vsnap, but then records blocks over time to create an exact copy.

3) Mirror Clone: Dynamic copy for DR, takes time for availability, but clone can be fractured, used then reattached.

While EMC does have a lot of granularity, the EVA was far more user-friendly. It enabled me to be more proficient at network storage operations, which we normally farm out, because of staffing constraints. The HP vs. EMC challenge was a great opportunity to really compare the products head-to-head and showed me the value of an easy to use SAN solution.

Posted By warrensander | 3 Comments | Trackbacks | Permalink
Filed under:


Comments

EVA Business Copy has always been a powerful and flexible tool. But the biggest problem with this particular tool is it's lack of a unified application integration suite. Yes I know HP AppRM and ZDB can do some of this, but overall application support is limited, all are complex to implement and functionality is well over a year behind Business Copy features. Give us a unified tool that will do SQL / Oracle / Exchange (including single mailbox restore) and has the option to snapback from a vsnap instead of just swapping clones. People are more interested in recovering from logical corruption than a physical LUN failure (that's why they buy EVA's and use raid). Give us a Netapp killer.
# Tuesday, March 04, 2008 11:33 AM by cleanur
AppRM is maturing in it's ISV integration capabilities with Oracle, SAP and Microsoft. With AppRM 6.1 which is scheduled to be released on March 31, Oracle and SAP recovery will be available for HP-UX and Windows environments for the fist time. AppRM currently supports Exchange and SQL recovery in Windows environments. Single mailbox recovery has always been available for Exchange recovery using Recovery Storage Group (RSG). RSG has the advantage that it does not change or affect the VSS shadowcopy – it is still pristine, read-only and available for use in fast recovery. Additionally, some of the plug-ins used by our competitors rely solely on snapshots, which impact the parent, instead of offering both space-efficient copies or fully-normalized snapclones. It is important to note that RSG is the only single mailbox method sanctioned by Microsoft. True, AppRM support has lagged Business Copy features, but try to remember that this software sits atop 3-5 other pieces (depending on the configuration) of HP software – there are a lot of pieces in the solution stack that have to come together for a perfect delivery. To combat delays in future deliveries, the AppRM team has taken steps to streamline the development process for more reliable delivery date communication and accomplishment.
# Tuesday, March 04, 2008 11:02 PM by karen.gilmer@hp.com
Thanks for the update and the for sharing some of the thinking behind using RSG and clones. Appreciate the performance hit, but there are ways to negate this somewhat now. Although I'm sure the capability to mount the copy to the RSG isn't or wasn't supported for clusters (vast majority of BC Exchange environments)due to identical disk sugnatures (MS problem). In this case having a mailbox data mining tool on the backup host makes a lot of sense.
# Wednesday, March 05, 2008 11:01 AM by cleanur

Leave a Comment

(required)  
(optional)
(required)  


Type the digits above:
Information disclosed in this community becomes public. Exercise caution when deciding to disclose your personal information. HP reserves the right, but is not obligated to, edit or remove your comment if it contains personally identifiable information or other content HP deems unacceptable.  Opinions expressed are your personal opinions or those of the original authors, and not of HP. Please see HP's web Terms of Use for more details.