RAID-DP

[Riley Martin]

Hi,

I've read about using RAID-DP when implementing Microsoft Exchange server and am unfamiliar with this type of RAID. Is RAID-DP known as any other name/RAID level? Does anyone know if any HP Proliant RAID controllers support this type of RAID. Thanks in advance for any help that anyone can provide.

Riley

[Rob Leadbeater]

Hi Riley,

RAID-DP is NetApp's implementation of RAID-6. See this technote:

http://www.netapp.com/library/tr/3298.pdf

On ProLiants you'll see it referred to as RAID-ADG or RAID-6.

Hope this helps,

Regards,

Rob

[Patrick Terlisten]

Hello Riley,

you shouldn't use RAID6 / ADG for Exchange, regardless if transaction logs or databases. RAID 3,4,5,DP,6 and ADG are suffering a phenomenon called "write penalty". This is the effort of a logical write IO. This one logical write IO is split into two physical read IOs (read block and read parity block) and two physical write IOs (write block and write parity block). Because RAID DP, ADG and 6 use two parity blocks, the effort is much bigger then RAID 3,4,5. Databases do often so called "short writes". A "long write" is the write of a complete stripe, mean all blocks in a stripe are changed and a complete new parity block is written. A "long write" is a little bit faster then a short write.

You shold use RAID 1+0 or RAID 1 for Exchange databases and transaction logs.

Hope this helps.

Best regards,
Patrick

[raadek]

RAID10 recommendation for random writes is fully valid in the world of traditional arrays (say MSA).

Yet the parity write penalty for short writes is not true in NetApp world - they do quite a simple trick, i.e. collect in NVRAM a bunch of short writes & flush them to disks in one go as one long write.

Regards

[Riley Martin]

Thank you Patrick, Rob for the information.

I'm a little confused however. Someone from an Exchange newsgroup gave me the below information but it's the opposite of what you said about using RAID-DP. Could you please read what they said below and let me know what you think? Thank you for helping me to understand which way to go with RAID and Exchange.

"With RAID 5, you get half the write throughput that you would with the same
number of spindles in RAID 10. In RAID 10 you get half the write throughput
of the same number of spindles in RAID DP. For Exchange 2003 with cached
clients, you can expect to see a read/write ratio around 2:1. A full third
of IO operations are writes. Yes, it makes a big difference in spindle
count to overcome the write penalty. If you concentrate only on space, and
ignore performance, then you need to prepare your resume and start looking
for a new job. Your implementation will be plagued by performance issues
and doomed to failure.

The general rule of thumb I use is: If the write penalty for a given RAID
type is higher than the read/write ratio of your application, then that RAID
type is inappropriate for your application.

With a write penalty of 4, RAID 5 is not appropriate for Exchange 2003.
With a write penalty of 2, RAID 1/10 would work for Exchange 2003.
With a write penalty of 1 (no write penalty), RAID DP is the best option
for Exchange 2003 from a performance perspective."

[raadek]

Riley,

This is actually a marketing discussion, i.e. judging which vendor technology is better.

On HP you have a choice between RAID10, RAID5 & RAID6 (on traditional arrays) or VRAID1 & VRAID5 (on EVA).

Definitely of these RAID10 or VRAID1 will give you the best performance in HP world.

On NetApp you basically have no choice performance-wise, but you can choose between either single parity (RAID4) or dual parity (RAID-DP).

So the guys you are quoting are simply pitching NetApp RAID-DP vs. some other vendor (e.g. HP) RAID10 or RAID 5 solution.

I believe this discussion forum is not the best place for such judgements.

Rgds.

[Patrick Terlisten]

Hello,

I can confirm Raadek's statement about RAID-DP.

But this forum is not the place for marketing statements. :)

Use RAID-DP on a Filer for Exchange, and RAID 1+0 or VRAID 1 on a HP Storage.

Best regards,
Patrick

[Riley Martin]

Thanks to everyone for clearing that up. For larger dedicated email servers I'll look into RAID 1+0 as an option and compare that against simply using RAID 1.

For small remote sites with 100 or less users I'm sure RAID 5 would be fine.

[Patrick Terlisten]

Hello,

don't forget to forgive points. :)

Best regards,
Patrick

[TTr]

Riley,

Although it is true about the write penalties of RAID-6,5,4,3 some of the new RAID products have reduced the write penalty to a point where a RAID5 or 6 LUN will outperform RAID-10. The breakpoint for that usually depends on the particular hardware and the number of spindles in your LUN. If you have the flexibility, do some testing by creating RAID10 and RAID6 or RAID5 LUNs and see for yourself what works best for you. For an accurate comparison you have to keep the LUN capacity the same across different RAID types and use the same spread of your spindles across the I/O buses on your storage system. So with 150GB spindles, if you use 8 spindles for RAID6, you have to use 12 spindles for RAID10 to get the same capacity and you have to spread the spindles the same way on you storage system's I/O.

[raadek]

TTr,

I am not saying there is no single vendor out there being able to overcome (or minimise) write penalty issues for RAID5/6 (NetApp being one of the examples).

Yet in the HP world we are looking at either SmartArray controller, MSA or EVA. In all these cases short writes on RAID5/6 get a significant performance hit due to underlying mechanism well explained above by Patrick.

Even on mighty EVA the best practice published by HP folks says "use VRAID1 for performance".

I've done a number of performance tests on traditional arrays comparing RAID10 with RAID5 (same number of spindles) & the former one just trashes the latter.

It's just the maths - to write a single block on a (traditional) RAID5 set, you have to read the parity info, calculate new parity, write parity & eventually write the actual data.

Rgds.

[Joshua Small_2]

Whilst RAID1+0 can be demonstratably faster in a test environment than RAID5, it's dangerous to take this at face value.

I've seen one group that purchased domestic SATA drives for a server, just so they could afford to use RAID1+0, because they read somewhere that it was "better", and noone could consider that the comparably priced RAID5 solution involving 10k SAS disks would be better.

I've worked with a number of slowly performing Exchange servers and it usually comes down to other services on the machine. Groupshield on your Exchange server has a terrible performance impact for example.

And if your deployment is big enough, Microsoft will suggest you deploy your server roles on different servers. Feeling you can avoid that be using RAID1+0 will definitely be wrong.

Only after you've considered the above, and the likelihood of networking (search these forums for performance issues with Hp server NICs at the moment), memory or similar issues should you even concern yourself with the whole RAID1+0 vs RAID5 arguement.

[Patrick Terlisten]

Hello,

don't forget that NetApp is using a simple trick: They cache write IOs in a NVRAM, so that they only write complete RAID stripes. A short write is often handled completely in the NVRAM. For security NetApp writes the NVRAM periodical to disks. They eleminate the write penalty simple by using a NVRAM.

This method could easily added by HP, but they don't do it. This would speed up every RAID level. But IMHO it's unsecure to do it all in a NVRAM.

btw: That RAID-DP is slow, while it's based on RAID 4 is one of the "silver bullets" which is used for marketing against NetApp. ;)

To make this story short: Use RAID-DP if you have a Filer. Use RAID 1+0 oder VRAID 5 if you have an MSA or EVA.

Best regards,
Patrick

[Riley Martin]

Joshua,

I'm new to Exchange. Could you explain further on Microsoft wanting server roles on different servers in a big deployment. Which roles, how big? Thanks.

[raadek]

Riley,

In Exchange 2007 you can have the whole bunch of roles: http://technet.microsoft.com/en-us/library/bb124935(EXCHG.80).aspx

In 2003 it is just about front-end/back-end split: http://technet.microsoft.com/en-us/library/aa996980(EXCHG.65).aspx

Patrick - you are spot on with the risk associated with NetApp NVRAM caching; that's why they mirror its content in the active-active config (it means using only half of the memory for actual caching).

Rgds.

[Joshua Small_2]

Being new, the real question to ask is, how big is your deployment?

If you have 20 users for example, I would suggest most of this thread is redundant, and complicating things by involving RAID levels will jsut confuse the issue.

If you have 10,000 users, there is far more to consider than what has been raised.