Couldn't agree more. Most of the users perception of the application performance is derived from the UI rendering. With the E8/9 five tier XML UI stack and the Infragistics tools lumbering about, it places a premium on CPU clock speed, backplane buss and memory performance - all of which have been vastly improved in newer workstation configurations.
I'm not at all surprised about your relative performance numbers on the server side. All of this talk of spindle counts and raid configurations ignores the fact that a well configured server should have most, if not all of the DB cached into RAM. As a matter of practice if you're not getting cache hit ratios well in excess of 95% your system is starved for RAM. If you are getting a reasonable cache hit ratio, then your I/O channel is mostly dormant. As a result, it makes perfect sense that 15% is the measurable difference between the two configurations. It probably has as much to do with the CPU/Memory differences as it does the magnetic I/O.
Then there is the case for SSD's. A handful of relatively inexpensive SSD's will significantly outperform a huge array of magnetic drives if only for the near zero latency they exhibit.
Michael Barry
Aspacia Systems Inc
866.566.9600
312.803.0730 fax
http://www.aspacia.com/
I'm not at all surprised about your relative performance numbers on the server side. All of this talk of spindle counts and raid configurations ignores the fact that a well configured server should have most, if not all of the DB cached into RAM. As a matter of practice if you're not getting cache hit ratios well in excess of 95% your system is starved for RAM. If you are getting a reasonable cache hit ratio, then your I/O channel is mostly dormant. As a result, it makes perfect sense that 15% is the measurable difference between the two configurations. It probably has as much to do with the CPU/Memory differences as it does the magnetic I/O.
Then there is the case for SSD's. A handful of relatively inexpensive SSD's will significantly outperform a huge array of magnetic drives if only for the near zero latency they exhibit.
Michael Barry
Aspacia Systems Inc
866.566.9600
312.803.0730 fax
http://www.aspacia.com/
On Jun 10, 2011, at 5:17 AM, Mordechai wrote:
> We have done extensive performance tuning and I feel that interactive performance it usually good on our systems.
> After reviewing everything from server memory to CPU and storage configuration I can say that you should shift your focus from the server to the client configuration and memory.
> We find a significant performance boost when users are on `current' technology � 2-3 year old computers will perform much slower.
> Don't know how many of you installed Epicor's performance Diagnostic tools, but a quick study of application server and network on different environments shows only little performance difference.
> So I compare my production running on 15K drives in a SAN with many drives on the latest CPU to my test environment running on 7.2K SATA SAN on a 4 year old server and see about 15% performance boost � try a fast client computer and you will start smiling � focus on the end user!
>
> Motty
>
> --- In vantage@yahoogroups.com, Zac Jason Woodward <zac@...> wrote:
> >
> > More or less restating what others have said already.
> >
> > "Ideal" configuration for any MSSQL server config is as follows
> >
> > RAID 1 OS
> > RAID 10 DB
> > RAID 10 Logs and Pagefile.
> >
> > The main concern is the DB RAID 10, the more drives you can throw at it the better. As for RAM 32 Gig is a safe place to be, but this all dependant on the number of users. Should you determine that you need to split the App server from the DB I would encourage that you purchase 10 gig ethernet cards to connect the 2 servers together.
> >
> > "Zac" Jason Woodward
> > Network Administrator
> > Intermountain Electronics, Inc.
> > O: 877-544-2291
> > M: 435-820-6515
> > F: 435-637-9601
> > www.ie-corp.com
> >
> > Creating customer confidence through extraordinary service and experienced industry experts.
> >
> > From: vantage@yahoogroups.com [mailto:vantage@yahoogroups.com] On Behalf Of Bethany Rye
> > Sent: Thursday, June 09, 2011 2:36 PM
> > To: vantage@yahoogroups.com
> > Subject: RE: [Vantage] Server Sizing for Epicor 9.05
> >
> >
> >
> > We are on 9.04.506C - SQL with 50 full clients and 30 MES clients. We are
> > running 3 servers:
> >
> > Server 1 - Database Server (SQL 2005):
> >
> > 2 x quad core Xeon 3.16 GHz processors
> >
> > 32 GB RAM
> >
> > Windows Server 2008, Standard 64-bit, Including Hyper-V, Including 5 CALs
> >
> > 2 x 146 GB 15K SAS mirrored (array 1)
> >
> > 2 x 450 GB 15K SAS mirrored (array 2)
> >
> > MD1000 DAS with:
> >
> > 6 x 300 GB 15K SAS, RAID 10 (array 3)
> >
> > 6 x 300 GB 15K SAS, RAID 10 (array 4)
> >
> > 16X DVD-ROM
> >
> > GB Ethernet NICs
> >
> > Tower Chassis
> >
> > SQL Server 2008 (Standard Edition) with processor licensing for 2 CPUs (but
> > to be loaded with SQL Server 2005 software)
> >
> > Server 2 - Application Server:
> >
> > 2 x quad core Xeon 3.16 GHz processors
> >
> > 16 GB RAM
> >
> > Windows Server 2008, Standard 64-bit, Including Hyper-V, Including 5 CALs
> >
> > 2 x 146 GB SAS mirrored (array 1)
> >
> > 6 x 300 GB SAS, RAID 10 (array 2)
> >
> > GB Ethernet NICs
> >
> > Tower Chassis
> >
> > Server 3 - Terminal Server:
> >
> > 2 x quad core Xeon 3.16 GHz processors
> >
> > 16 GB RAM
> >
> > Windows Server 2008, Standard 64-bit, Including Hyper-V, Including 5 CALs
> >
> > 2 x 146 GB SAS mirrored (array 1)
> >
> > 6 x 300 GB SAS, RAID 10 (array 2)
> >
> > GB Ethernet NICs
> >
> > Tower Chassis
> >
> > Hope this helps!
> >
> > Beth Rye
> >
> > IT Director
> >
> > CIGNYS
> >
> > ***ITAR NOTICE***
> >
> > This e-mail and/or the attached documents may contain technical data within
> > the definition of the International Traffic in Arms regulations, and are
> > subject to the export control laws of the US Government. Transfer of this
> > data by any means to a foreign person, whether in the US or abroad, without
> > an export license or other approval from the US Department of State, is
> > prohibited. No portion of this e-mail or its attachment(s) may be reproduced
> > without written consent of CIGNYS. If you are not the intended recipient or
> > believe that you may have received this document in error, please notify the
> > sender and delete this e-mail and any attachments immediately.
> >
> > From: vantage@yahoogroups.com<mailto:vantage%40yahoogroups.com> [mailto:vantage@yahoogroups.com<mailto:vantage%40yahoogroups.com>] On Behalf Of
> > jplehr
> > Sent: Thursday, June 09, 2011 3:52 PM
> > To: vantage@yahoogroups.com<mailto:vantage%40yahoogroups.com>
> > Subject: [Vantage] Server Sizing for Epicor 9.05
> >
> > I am posting for a new user who does not have access to the group at this
> > time.
> >
> > Looking to get some feedback on the sizing of the hardware they purchased
> > and what Epicor is now telling them.
> >
> > Dell PowerEdge R710
> > Dueal E5620 2.4 GHz Xeon
> > 8 GB (they are bumping to 32 GB)
> > Raid 1: 2x 146 GB 15k SAS
> > Raid 10: 4x 600 GB 15k SAS (I believe they are getting pushed to add
> > another)
> > Windows Server 2008 Std. R2
> >
> > Will be running Epicor 9.05 SQL w/ Enterprise Search.
> >
> > Any thoughts from you E9 users.
> >
> > [Non-text portions of this message have been removed]
> >
> >
> >
> > [Non-text portions of this message have been removed]
> >
>
>
[Non-text portions of this message have been removed]