That I don't know. I believe they are Fujitsu Fibre Channel drives (10K and
15K) for our Clariion CX3-10.  Not sure on the OEM of the SATA drives for
the same.  For my application it's a SAN being used for Oracle, but I'll
double check those numbers.

Come to think of it, it was a lunch meeting with the engineer and our sales
guy, so he could have been exaggerating some.  :)

Thanks for the explanation.

Joe Fox
Systems/Network Administrator

Mobile# (716) 846-9308
http://www.linkedin.com/in/josephfoxjr


On Tue, Sep 23, 2008 at 11:19 AM, Michael B. Smith <
[EMAIL PROTECTED]> wrote:

>  Is that sequential read only?
>
>
>
> 'Cuz those numbers seem high. Very high. What are the physical drive types
> and models? If you can't get those, the relevant metrics are: 1] inter-track
> latency, 2] full-bore average read access, and 3] full-bore average write
> access.
>
>
>
> For Exchange 2007 database access, which is completely random, per 10 K RPM
> spindle you generally see something like 1] 1 ms, 2] 8 ms, and 3] 12 ms. On
> average, read:write is 1:1, so you average 8 + 12 = 20 / 2 = 10 ms average
> access. This gives you an IOPS of 1000 ms / (10 ms + 1 ms) = 91 IOPS
>
>
>
> Big difference in performance and capacity planning between 91 IOPS and 120
> IOPS.
>
>
>
> I covered this recently in:
>
>
>
>
> http://theessentialexchange.com/blogs/michael/archive/2008/09/19/it-s-all-about-the-iops-silly.aspx
>
>
>
> and
>
>
>
>
> http://peach.ease.lsoft.com/scripts/wa.exe?A2=ind0809&L=EMO-NEWSLETTER&T=0&F=&S=&P=1470
>
>
>
> Note that 100% sequential read eliminates the inter-track latency, and the
> same drive gives you an IOPS of 125, much closer to the numbers you were
> told. You need to verify that you are being told what you THINK you are
> being told. Salespeople tend to quote the most favorable number…
>
>
>
> Regards,
>
>
>
> Michael B. Smith, MCITP:SA,EMA/MCSE/Exchange MVP
>
> My blog: http://TheEssentialExchange.com/blogs/michael
>
> Link with me at: http://www.linkedin.com/in/theessentialexchange
>
>
>
> *From:* Joe Fox [mailto:[EMAIL PROTECTED]
> *Sent:* Tuesday, September 23, 2008 11:02 AM
>
> *To:* MS-Exchange Admin Issues
> *Subject:* Re: SAN Setup Recommendations
>
>
>
> I just met with my EMC Rep last week as we need to add my disk to our SAN.
> Here are the figures that he gave me on IOPS:
>
> 15K RPM - 180 IOPS (FC Drives)
> 10K RPM - 120 IOPS (FC Drives)
> 7.2K RPM - 70-80 IOPS (SATA Drives)
>
> HTH.
>
> Joe Fox
> Systems/Network Administrator
>
> Mobile# (716) 846-9308
> http://www.linkedin.com/in/josephfoxjr
>
>  On Wed, Sep 17, 2008 at 1:52 PM, Sean Martin <[EMAIL PROTECTED]>
> wrote:
>
> I'm not at all experienced on Exch 2007, but if the same disk
> recommendations for Exch 2003 still hold true, I have the following advice:
>
>
>
> 1) Migrate one of your transaction log LUNs to a separate, dedicated RAID 1
> group. You want to have your logs on separate spindles if at all possible.
> I'm not familiar with HP's SAN offering, but you should be able to migate
> the LUN and have it be transparent to the host. (Since you mentioned having
> 2 VRAID1 on the second disk group, I'm assuming the host see's these as two
> different physical disks.)
>
>
>
> 2) Typically RAID 10 is recommended for hosting info stores, but depending
> on your user load (read: IOPS), RAID 5 may suit your needs just fine.
> However, given the number of DBs you're supporting (and not knowing the
> number/type of users) I would be inclined to recommend at least two RAID5
> groups (on dedicated spindles) for hosting each of your storage groups.
>
>
>
> 3) It may be too late, but I believe it is recommended to build separate
> storage groups before populating each storage group with multiple databases:
>
>
>
> Your Setup:
>
> SG1 - DB1, DB2, DB3, DB4
>
> SG2 - DB5, DB6, DB7, DB8
>
>
>
> Recommended:
>
> SG1 - DB1, DB5
>
> SG2 - DB2, DB6
>
> SG3 - DB3, DB7
>
> SG4 - DB4, DB8
>
>
>
> The obvious disadvantage is you would want to provide 4 separate RAID 1
> groups for logs.
>
>
>
> - Sean
>
> On Wed, Sep 17, 2008 at 9:15 AM, Jeremy Phillips <
> [EMAIL PROTECTED]> wrote:
>
>  How many disks are in each disk group?
>
>
>
> Thanks,
>
> Jeremy Phillips
>
>
> -----Original Message-----
> From: Travis Krampy [mailto:[EMAIL PROTECTED]
>
> Sent: Wednesday, September 17, 2008 10:12 AM
> To: MS-Exchange Admin Issues
> Subject: Re: SAN Setup Recommendations
>
> There are 2 disk groups actually, one that is a RAID5 and one that is a
> RAID1
>
> The raid 5 disk group is split into 8 VRAID5 sets and the RAID1 disk group
> is split into 2 VRAID1 sets.
>
> Sorry I was not clear on this.
>
> Plus this is inherited, i never set this up, thats why im asking the
> experts!
>
> Travis
>
> "Barsodi.John" <[EMAIL PROTECTED]> wrote in message
> news:[EMAIL PROTECTED]
> Sounds like you have one disk group and you've created several LUNs with
> varying Vraid types(1,5) within that Disk Group.  Just remember the
> Vraid1 and Vraid5 sets are sharing the same disks within that Disk
> Group.  Log writing and DB writing have different write
> patterns(sequential vs. random) so placing them on the same set of
> spindles could cause head contention.  Best practice is separate
> spindles for logs and DB.   Sounds like resources are limited, and
> depending on your performance requirements, this could be fine.
>
> - John Barsodi
>
> -----Original Message-----
> From: Travis Krampy [mailto:[EMAIL PROTECTED]
> Sent: Wednesday, September 17, 2008 9:50 AM
> To: MS-Exchange Admin Issues
> Subject: Re: SAN Setup Recommendations
>
> I think, from how I am looking at it, it seems that all disks on the are
> all
> together in one big array (RAID5) then broken down into several smaller
> VRAID5 arrays.
>
> It looks like there is about 2TB of disk space allocated to Exchange
> each
> are 72Gb 10k drives
>
> Does that help any?
>
> Thanks
>
> Travis
>
>
> "Martin Blackstone" <[EMAIL PROTECTED]> wrote in message
> news:[EMAIL PROTECTED]
> > I'm not sure I understand it either.
> > Are these luns on separate spindles or all the same ones? How many
> disks
> > are
> > there dedicated to Exchange?
> >
> > -----Original Message-----
> > From: Jeremy Phillips [mailto:[EMAIL PROTECTED]
> > Sent: Wednesday, September 17, 2008 9:01 AM
> > To: MS-Exchange Admin Issues
> > Subject: RE: SAN Setup Recommendations
> >
> > Sounds right at first glance (I'm not sure I understand *exactly* how
> the
> > SAN is configured) but my main question would be whether the drives
> can
> > support the iops needed?
> >
> > Thanks,
> >
> > Jeremy Phillips
> >
> > -----Original Message-----
> > From: Travis Krampy [mailto:[EMAIL PROTECTED]
> > Sent: Wednesday, September 17, 2008 8:28 AM
> > To: MS-Exchange Admin Issues
> > Subject: SAN Setup Recommendations
> >
> > Hi All,
> >
> > Forgive me if I am not explaining this correctly...
> >
> > I am currently working in implementing an Exchange 2007 CAS/HT server
> as
> > well as an exchange 2007 server as a secondary mailbox server with the
> > existing Exchange 2003 seerver.
> >
> > Currently the Exchange 2003 server is setup to have all Logs and DB's
> on a
> > SAN.
> >
> > Before I set up the new exchange 2007 back end server, i want to make
> sure
> > that the SAN is setup properly for Exchange.
> >
> > As of now, there are 2 storage groups with 4 stores in each storage
> group.
> >
> > On the SAN, each store is stored in 8 individual Luns that are VRAID 5
> >
> > the logs are stored seperately for each storage group in their own LUN
>
> > that
> > is a VRAID 1
> >
> > so on my exchange server, i have 8 drives that are for stores, and
> drives
> > for logs.
> >
> > Does this sound like the SAN is implemented correctly for Exchange?
> >
> > Please advise
> >
> > Thanks
> >
> > Travis
>
>
>
>
>
>
>

~ Ninja Email Security with Cloudmark Spam Engine Gets Image Spam ~
~             http://www.sunbeltsoftware.com/Ninja                ~

Reply via email to