Showing posts with label SBS Hardware. Show all posts
Showing posts with label SBS Hardware. Show all posts

Thursday, 28 April 2011

SAS versus SATA and Hardware RAID versus Software RAID

In the last few years we have made some changes to the server configurations that we either build and deploy or Tier 1 provides:

  • We install SAS 10K or SAS 15K drives over SATA.
    • Performance for one is vastly superior on the SAS drives.
    • SAS drives are a lot more sensitive to bad sector behaviour and are better able to recover from bad data being tossed up.
    • SAS drives use many forms of ECC which protects the integrity of the data.
  • We install a hardware RAID controller with battery backup or SSD Cache over using the on board software RAID.
    • The on board “RAID” is software driven. All RAID calculations are completed by the server’s CPU and in many cases require the software driver to rebuild or function properly – meaning we need to boot into the OS.
    • We have had difficulties with on board software RAID recoveries and ShadowProtect due to driver issues.
    • Hardware RAID on Chip with the battery backup or SSD Cache virtually eliminates the parity write cost of RAID 5 and greatly reduces that cost for RAID 6. SSD Cache almost renders the whole discussion moot as the most frequently requested data sits on the SSD.
    • Hardware RAID controllers have the ability to mitigate the failure of a drive by keeping the server up. In our experience software RAID tends to freeze the box if a drive fails.
    • Hardware RAID controllers tied to SAS drives have a much better chance of mitigating or eliminating the possibility of data corruption if sectors on an array member are dying.

Intel has a couple of resources for the above discussion as do many other RAID related vendors.

One Intel resource is the following Intel support page: Intel Server Products: Choosing between SAS vs. SATA Hard Disk for your Server RAID System.

There we find this grid that gives us a pretty good idea of some of the significant differences between SAS and SATA drives:

image

We are given an extensive explanation in the following document which is linked to below the above table: Intel – Enterprise-class versus Desktop-class Hard Drives (Link to PDF document download).

We put our server configurations through a lot of testing before we deploy them to client sites or within our own organization. We do this because we want to make absolutely sure that the server configuration we are going to deploy will meet the needs of our client over the life of the box which is about 36 months.

The extra cost for the hardware RAID controller, battery backup, and 15K SAS drives (not much of a cost difference between 300GB 15K SAS and Seagate Enterprise SATA these days) when taken over the life of the box (divide that cost by 36) is actually quite small relative to the performance, data protection, and overall storage stability benefits.

Tier 1 Caveat

Keep in mind that cost inferences mentioned here are for our in-house Intel server solution components. When it comes to Tier 1 the costs of some server components and server storage can be extremely high relative to components supplied in our solutions.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*Our original iMac was stolen (previous blog post). We now have a new MacBook Pro courtesy of Vlad Mazek, owner of OWN.

Windows Live Writer

Wednesday, 23 March 2011

Physical and Virtual Server Configuration Norms For Us

The following is one of our responses in a hardware configuration conversation for SBS 2011:

***

. . . it really does get difficult to work through all of this for those that may not spend time working directly with hardware.

Since we build our own servers and have been doing so for a good 12-15 years now we know the hardware side of things quite well.

We also resell tier 1 for remote clients though we don’t like the $*?$! match between them and software vendors when things break. Again, given our hardware experiences we can bring both in line most of the time and get a fix.

That being said, experience tells us that as a general rule:

  • For virtualized environments more sockets + cores is better.
    • All VM threads are processed in parallel for 2+ vCPUs.
    • Hex-Core CPUs hit a sweet spot for virtualized setups where 4 vCPUs are assigned to several VMs.
  • For physical environments more GHz is better.
    • We use Intel Xeon CPUs exclusively. YMMV.
  • More RAM is always better.
  • Hardware RAID is a must.
    • More spindles is better.
    • PCI-E 8x 2nd Gen is better.
    • Battery Backup or SSD Cache is better.
  • 15K 2.5” spindles is better than 15K 3.5” spindles.
    • SATA is only Nearline storage and never for primary production I/O.
  • SSD blows away any and all spindle based setups.

That about sums up the overall perspective we have on server hardware.

***

Using the above guide we are able to start with a base platform that is tailored to a client’s needs and then scale up from there.

In addition to the above, we can add redundancy features such as multi-port server NICs, additional power supplies, clustering configurations, and more.

In the end it is our aim to provide the best all around IT solution for our client’s needs.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*Our original iMac was stolen (previous blog post). We now have a new MacBook Pro courtesy of Vlad Mazek, owner of OWN.

Windows Live Writer

Tuesday, 18 January 2011

8GB Intel Validated Kingston Memory For S3420GP(RX) – Finally!

When it comes to supporting Intel’s server products, one of the first questions we get from the support technician when we start troubleshooting a problem is, “What type of memory is in the system?”

If that memory is not on the tested memory products list for that particular server board then we will be asked to install Intel Validated memory products.

Up to now there have been two 8GB sticks available for the Intel Server Board S3420GP(RX), one by Hynix and the other by Samsung (though not listed now), listed as validated for this series of server boards and the server systems based on them.

That has finally changed!

We now have two new Intel Validated Kingston parts:

  • KVR1333D3D4R9S/8Gi
    • Cost: ~$350/Stick
  • KVR1066D3Q8R7S/8Gi
    • Cost: ~$200/Stick

The catch for us here in Canada is that there is lots of stock for the 1333 MHz part while the 1066 MHz parts are in short supply at the moment. There are also kits of 2 (16GB) for each part number as well.

Why is this so important?

Because the Hynix and Micron parts have only been available via resellers in the US. So, any warranty issues were questionable since that list of resellers is very small and they are relatively unknown to us. RMA processes can be painful with some manufacturers and their resellers so we are cautious when it comes to new points of product supply.

We have had a lot of success with Kingston’s memory parts and when there have been times where we have needed to call them they have always been very helpful.

So, this news is a definite plus for us!

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*Our original iMac was stolen (previous blog post). We now have a new MacBook Pro courtesy of Vlad Mazek, owner of OWN.

Windows Live Writer

Friday, 3 September 2010

SBS 2008 Entry Level Server Configuration

Here is a sample Intel based server that we would configure for a small SMB with 5-10 seats.

One can find a similar setup in a Tier 1 machine, though the cost may be a bit higher.

Essentially, today’s entry level server systems that have been configured _properly_ to run Small Business Server 2008 should run about $3,000.00 (in Canadian Dollars).

Now, take that $3K and divide that over the life expectancy of the server:

  • $3K/3 = $1,000.00/Year

Okay, now, let’s say we are dealing with a client that has 5 seats.

  • $1K/5 = $200/Year/User

Now, the next step is to break that figure down a little more to provide some context to the business owner:

  • $200/1800Hrs = $0.11/Hr per user
    • The 1800Hrs per year is calculated roughly on 50 weeks worked with some days off for statutory holidays.

Part of the discussion is to also ask our contact for a general gross annual income for the company. We take that number and break it down just as we do the costs. We generally end up with an employee’s value being anywhere from $65/Hour all the way up to $125/Hour.

Knowing these details helps to explain the value proposition of our IT solution.

During our conversations we spend a fair amount of time learning what our clients or prospective clients value in their IT solutions. That can be anything from user’s productivity to having a disaster recovery plan and the appropriate backup structures in place.

When we know what our contact values the most in their IT solution, we can place a dollar figure on it and start talking to them about the return on their investment (ROI) versus their total cost of ownership (TCO) without using all of those fancy acronyms or Geek Speak.

Explained this way, the total solution cost starts to take on a _very_ different perspective.

Have a great Labour Day Weekend everyone and thanks for reading! :)

UPDATE: Added a sample configuration sheet to our Intel Server Configuration downloads page.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*Our original iMac was stolen (previous blog post). We now have a new MacBook Pro courtesy of Vlad Mazek, owner of OWN.

Windows Live Writer

Wednesday, 28 July 2010

SBS v7 Memory and Resource Usage

Here are some screenshots for you:

image

Yes, that is the Exchange store weighing in at a comfortable 2GB+ of RAM!

The box itself is not too busy right now, but there will need to be some serious evaluation done on the purposing of the SBS box beyond running SBS and perhaps an AntiVirus management console.

image

This box is an Intel SR1530HSH Server System running an Intel S3200SH server board with 8GB of RAM. An RS2BL040 RAID controller and three 450GB 15K.7 Seagate SAS drives round out the configuration in RAID 5 as there are only three 3.5” hot swappable drives in this rig.

Based on our initial observations any server needing to run SBS v7 should be at least Quad Core preferably with HyperThreading for a total of 16 threads. It should have 16GB of RAM and preferably have a hardware accelerated RAID controller along with SAS drives.

Since we are a relatively small shop, we will run with the RAID 5 setup on this box until we migrate to the next version of SBS v7. Then we will purpose a pedestal box with a bit more RAM and some extra drives in RAID 10 to see if we notice any performance differences.

We will be shuffling the SharePoint Foundation content database off of SBS v7 and onto a dedicated SQL box to save on some of the migration time for SharePoint. It will also save a bit on the disk I/O and RAM as we use SharePoint _a lot_ in our shop and via the Internet.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*Our original iMac was stolen (previous blog post). We now have a new MacBook Pro courtesy of Vlad Mazek, owner of OWN.

Windows Live Writer

Wednesday, 21 July 2010

Beer Budget Hyper-V and/or Lab Server

Sometimes past memories happen to drop in when a question gets asked or something happens to trigger them.

One of those moments came when looking at a question we answered for someone going way back where they asked how to build a beer budget lab box.

At the time, server based hardware was significantly more expensive than the entry grade server equipment that we can get today.

So, in answering this question, the memory that popped up was one of the first lab boxes that we ever had. It was a server board that was a few generations prior to current, a pair of Intel CPUs running at what speed we are not sure of (not Tualatin CPUs), 32MB or 64MB RAM, and a pair of IDE or SCSI hard drives. Everything was sitting essentially out in the open.

Eventually they were mounted to the lid of an old desktop case much like our current Data Mule system was.

 image

MPECS’ Data Mule system

Keep in mind that there is a need to know how to use a Tap & Die set to create the threaded bores the motherboard standoffs will need to be set into if custom mounting the server board to an old case shell.

Otherwise, an old Styrofoam hard drive shipping container will work fine for setting the server board and drives into for protection but minimal initial cost.

So, here is the beer budget server setup that will allow for Hyper-V to run with hardware virtualization with prices showing in Canadian dollars.

  • Intel S3420GPLC $232
  • Intel Xeon X3430 $209
  • RAM configuration options
    • 8GB (2x 4GB) Kingston $290
    • 16GB (4x 4GB) Kingston $580
  • 2 SATA drives RAID 0 $120
    • no redundancy but gives performance.
  • Decent three rail 550 Watt PSU $80
  • OPTIONS
    • Intel PRO series dual gigabit NIC $200
    • Intel SC5650UP server chassis $250
    • 16GB (2x 8GB) Hynix $840
    • 32GB (4x 8GB) Hynix $1,680
    • Intel RS2BL040 PCI-E RAID $350

Using the above guide, we can have the bare minimum lab box up and running with 8GB of RAM for $931 plus applicable taxes and/or shipping charges if needed.

We all have USB keyboards and mice kicking around or can be had for pennies. Plus, a local system builder would probably be more than happy to part with a few bags of system screws, standoffs, and the like for a couple of bucks.

A DVD optical drive can be had very easily though newer OS installs can be done via USB flash drive thus eliminating the cost there.

As the budget allows, speed up those installs by purchasing faster 8GB USB flash drives with the best speeds being a real-world 35MB/Sec read and 30MB/Sec write. OCZ makes some great ones in the Rally2 Turbo and the ATV Turbo.

Note that USB 2.0 flash drives will become harder and harder to find as USB 3.0 drives come in. So far, the USB 3.0 drives are _a lot_ more expensive than their older USB 2.0 siblings.

BTW, a couple of LED fans would provide stylish cooling for that extra touch . :D

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*Our original iMac was stolen (previous blog post). We now have a new MacBook Pro courtesy of Vlad Mazek, owner of OWN.

Windows Live Writer

Wednesday, 16 June 2010

USB 3.0 – Is It Worth It?

We are gearing up to do some testing here in the shop of some new USB 3.0 gear.

The components:

We will use a standard 7200 RPM Seagate SATA drive to provide the baseline for performance. We will then use a Western Digital 150GB VelociRaptor and an Intel 80GB G2 code SSD drive to push the USB 3.0 bus to its limits.

We already know that USB 2.0 gives us a real world throughput of about 35MB/second.

We have seen eSATA give us around 60-70MB/second in our own real world testing and usage of the connection.

So, we look forward to seeing if USB 3.0 is all its cut out to be.

Once we have a good idea of what the speed advantages are, we will install the Vantec USB 3.0 PCI-E Host Card into one of our lab servers with an SBS 2008 OS installed. We will configure the SBS Backup to run once an hour. Hopefully we will be able to gain some insight as to whether the Vantec card will work with our servers.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*Our original iMac was stolen (previous blog post). We now have a new MacBook Pro courtesy of Vlad Mazek, owner of OWN.

Windows Live Writer

Saturday, 22 May 2010

SBS 2008 Install – No NIC Found For Xeon Processor 5600 Or Xeon 3400 Series Servers

We have found that SBS 2008 RTM or SBS 2008 with SP2 slipstreamed into the media will not find some of the newer Intel based NICs in our newer Xeon 5500/5600 and Xeon 3400 Series servers.

On one of our recent Answer File installs we were greeted with the following screen the next day:

image

Install Windows Small Business Server 2008

A network adapter was not found

We plugged in a USB flash drive that had the most current Intel ProSet driver executable on it.

We then clicked the Browse link shown in the screenshot above.

We navigated to the USB flash drive and double clicked on the ProSet file. Once the driver install routine spooled up, we made sure to select only the NIC Driver.

We were then saw:

image

Once the NIC drivers were set our SBS install routine continued on from there:

image

When we have a bit of time we will use the Windows Automated Installation Kit (Win 7) to inject the new Intel Pro Set drivers into the SBS 2008 SP2 WIM and burn a new DVD thus eliminating any further install interruptions.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*Our original iMac was stolen (previous blog post). We now have a new MacBook Pro courtesy of Vlad Mazek, owner of OWN.

Windows Live Writer

Thursday, 25 June 2009

Dell Default Server Partitioning Scheme – SBS Installers Beware

This is the default partitioning scheme that we encountered via a support call with someone having problems with their SBS 2008 install:

image

If the factory OEM install was used, or if the Dell preparation utility disk was used and the utility was allowed to install the Dell Utility Partition, the above is what we would end up with.

In our case, we always flatten the RAID arrays and start fresh without the utility partitioning schemes and set up our default partitioning for SBS 2008 (SBS 2008 Setup Checklist post).

The catch with the setup above is that the SBS 2008 Backup will fail out of the box. That FAT32 partition will break it.

Using the Diskpart command we can convert that partition to NTFS and at least get the SBS 2008 Backup to run successfully.

In the end, the above leads to a messy partitioning scheme along with the possibility of causing problems in the event of a failed RAID array.

We prefer to keep things simple. :)

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*All Mac on SBS posts will not be written on a Mac until we replace our now missing iMac! (previous blog post)

Windows Live Writer

Monday, 11 May 2009

Allowing a Dell DRAC (Remote Management) Session through ISA

The Dell Remote Access Card (DRAC) requires a couple of ports to be opened up outbound through our ISA servers in order for the Console redirection function to work correctly on a remotely managed server:

  • TCP Outbound: 5900 (Keyboard and Mouse)
  • TCP Outbound: 5901 (Video)

Create a custom protocol with the above settings, we call it Dell DRAC, and then create an Access Rule that allows the custom protocol from Internal/Local Host to External for All Users.

image

Once the rule has been created we are able to remote into any Dell server that has a DRAC 5 installed.

For any client that we support remotely, we make sure to have a remote management capability in the box ahead of time. This gives us console access via an Internet connection for those times where an OS may have stopped responding and we need to power cycle the box. Or, for SBS updates that like to kill Exchange, IIS, and/or RRAS thus taking us out of our remote session via the Remote Web Workplace or PPTP VPN.

As a result, we no longer need to have a timed script in place to force the box into a reboot during updates which may not bring the box back up in some cases where an update chokes the NICs, RRAS, or even IIS.

An additional benefit to having a remote management capability is being able to watch the boot cycle from start to finish. We can see all of the BIOS, RAID controller, and other firmware messages on our screen prior to the OS loading. It is one more way for us to assess the health of the server.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*All Mac on SBS posts will not be written on a Mac until we replace our now missing iMac! (previous blog post)

Windows Live Writer

Tuesday, 14 April 2009

LSI Corp to purchase 3Ware/AMCC Storage Adapter Business

In an interesting twist of events:

Intel, Dell, and other manufacturers use LSI boards as the foundation for their RAID controller offerings.

How this purchase pans out for the RAID controller market has yet to be determined. We do hope that the 3ware products live on since they are as good as the Intel/LSI products we deal with and competition is a healthy thing for manufacturers and for our clients and us.

For us, Intel/LSI are the RAID controllers of choice. We have had great success with Intel’s cards over the years. So, we will continue to stick with them.

The only other RAID controller we would look at for specialized applications would be by Adaptec. Though that need has been virtually nonexistent for a few years now.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*All Mac on SBS posts will not be written on a Mac until we replace our now missing iMac! (previous blog post)

Windows Live Writer

Tuesday, 24 March 2009

Upgraded Recovery and Migration Test Server

One of the benefits of providing Intel based hardware is the ability to recycle their parts, or the entire box depending on our requirements for a server.

As part of most client server upgrades we remove any of the old hardware from the client’s site. Depending on the hardware configuration, the box will end up in our phenomenal Alberta Electronics Recycling Program, or put into some noncritical service needs such as our lab.

This screenshot is the result of combining components from two servers we recently upgraded at client sites:

09-03-24 Intel Pentium D Recovery and Migration Server

Intel Pentium D server with RAID and hot swap

The internals came from one server and the chassis from another.

This particular box will replace our older recovery test server that was used to provide our regional and remote clients with their quarterly backup recovery testing services.

Now, we will be able to use this particular box to recover not only our existing SBS 2003 clients using ShadowProtect, we will be able to use the Hardware Independent Restore capabilities in SP as well due to the newer hardware on this box.

This “new” unit will give us the ability to restore the larger server configurations that are running either SBS 2008 Backup or Windows Server 2008 x64 Backup without worrying about storage limitations (3TB in some cases) or hardware capacity problems. We will set the drives to a RAID 0 stripe to reach that level of data volume. The added benefit is a performance boost with striping the array across the six spindles.

Having this new box gives us the ability to run SBS 2003 to SBS 2008 migration tests for our client’s production SBS servers using their newly configured server box along side of this one. It is our preference to run the two SBS OSs on hardware when possible. Another benefit to the “new” lab box is the hardware is actually quite similar to our clients existing SBS 2003 server configurations!

And finally, the RAID controller in this box is an Intel SRCS16 PCI-X. We will be able to now start our Hardware Independent Restore testing using the built-in SBS 2008 Backup as well as the Windows Server 2008 Backup!

The box configuration:

An external USB DVDRW is used to provide optical media access for all of our lab based boxes.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*All Mac on SBS posts will not be written on a Mac until we replace our now missing iMac! (previous blog post)

Windows Live Writer

SBS 2008 – Using SBS Backup to Expand an Array and Partition

When we set up our SBS 2008 installs, we use a RAID 1+0 configuration with three partitions on the RAID array.

We do this because of the extra throughput we get for both reads and writes. RAID 5 takes a pretty good performance hit on writes, while reads are similar to a RAID 0 stripe.

Our partitioning is done as follows:

  • System C: 100GB
  • Swap File S: 25GB (2.5 * RAM)
  • Data Partition L: GB Balance

Now, since we do keep the Exchange databases on the system partition, the size of the organization is an important consideration when it comes to the size of it.

For some RAID controllers, there is the ability to add disks to the array on the fly. From there, we can open the Disk Manager and Extend the last partition on that array.

To resize, move, or modify any of the other partitions with the newly acquired space will require a third party utility that can work with server partitions.

In the case where the RAID array cannot be expanded on the fly, we have the built-in SBS backup to use as an intermediary. The catch is the need to delete the existing array and create a newer and larger one in the RAID controller’s BIOS.

The whole process is actually quite simple and will take a few hours depending on the size of the backup being used.

  1. We take a backup just prior to downing the server.
  2. Shut down the server.
  3. Insert the new drives.
  4. Reconfigure the RAID array by deleting the existing one and creating a new one incorporating the new disks.
  5. Boot from SBS 2008 DVD 1.
  6. Click the Repair link.
  7. Restore my PC.
  8. Choose the backup just made for the restore.
  9. Make sure to load the RAID drivers just in case during the backup restoration steps.
  10. Run the restore.
  11. Go have coffee, tea, or take on another short project.

The restore process is set to automatically reboot into the newly restored OS once it finishes.

From there, head into the Disk Manager in the SBS Native Tools Management Console and extend that Data partition.

The new SBS 2008 Backup is phenomenal. The more we experiment with its abilities, the more we grow in confidence using it as our default backup program.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*All Mac on SBS posts will not be written on a Mac until we replace our now missing iMac! (previous blog post)

Windows Live Writer

Monday, 23 March 2009

On Board RAID vs. Add-In RAID Controller and Drive Failure Behaviours

When it comes to balancing the abilities of a server versus its cost, one of the first things to go is the add-in RAID controller.

When it comes to the upfront cost for the add-in RAID controller, it is virtually paid for in improved performance.

The catch with that though is when the first drive failure happens. While not common, drives are mechanical in nature and do fail over time.

The behaviour we have seen on an array member failure when connected to the on board RAID is a server lock-up. Now, for SBS, we can catch it pretty quick when we do not receive a report on the server in the morning, or via phone call if the failure happens during the business day.

There is also the additional cost of having us come down in emergency mode to make sure everything comes back up okay if there was a lock-up.

The add-in RAID controller on the other hand does not flinch. It has the ability to work with an array member failure and transitions the array state accordingly without impact to the server.

While some of the newer on board RAID controllers have the ability to do hot spares, the rebuild of the hot spare into an existing array with a failed member will directly impact the server’s performance.

A hot spare that is dropped into a failed array member’s spot on an add-in RAID controller will not impact the server’s performance for the most part.

As a result of some of the drive failure situations that we have seen, we no longer forgo the add-in RAID controller in any of our client’s sites. It does not pay in the long run.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists
Co-Author: SBS 2008 Blueprint Book

*All Mac on SBS posts will not be written on a Mac until we replace our now missing iMac! (previous blog post)

Windows Live Writer

Tuesday, 9 December 2008

Canadian and Other Non-US Currency Quoting Cautions

With the economies of the world struggling and in some cases dropping in the tank, we have seen a roller coaster ride of ups and downs for our product pricing.

This is what our Canadian dollar looks like against the United States dollar today from XE:


The Canadian Dollar Today

Ouch. :(

Prices in the distribution channel have only started increasing in the last two to four weeks for us. In some cases, we received price increase warning letters from manufacturers that we are registered resellers for. In others, Synnex Canada has been pretty good about e-mailing us with price increase warnings.

Pricing volatility has played havoc on our quoting which we normally allow 30 days for. Lately, we have reduced that to one business week.

As a result, we end up needing to buffer the quote up 5-10% (divide not multiply) to cover any immediate increases.

If prices remain relatively static, then when the time comes to issue an invoice for the deposit on the project, we will let our client know that we can pass along a 5-10% savings to them due to our pricing levels.

The tough spot was about three weeks ago when we saw the Intel Xeon E5440 Quad Core CPUs jump by close to $300 cost per processor and the RAID controller jumping another $150 in cost. Other components in the server build also saw some significant steps upwards in price.

We were able to talk to our clients and prospective clients about a compromise on our existing quotes. They understood that the economic situation was making things difficult for us. We ended up absorbing 50% of the cost increase for that situation. Another ouch, but at least they were happy that we took the time to explain the situation and split the difference with them.

For those of you in the United States, this situation may not be impacting your business, but for those of us outside of the U.S., we are, in some cases, being hit quite hard with the currency fluctuations.

Be cautious in quoting folks. Given the situation in the world, now is not the time to start losing on projects due to missed margins.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists

*All Mac on SBS posts are posted on our in-house iMac via the Safari Web browser.

Thursday, 26 June 2008

Seagate 750GB ES.2 Failure on SR1530AHLX swap experience

This is our second drive failure for this series within a short period of time.

The drive:
  • 750GB ST3750330NS Seagate ES.2
    • Firmware: SN04
    • Date Code: 08281
    • Site Code: KRATSG
    • Product of Thailand
In this case, there were two drives connected to the S3000AHLX on board RAID controller in a RAID 1 array.

The first indication of a problem was the lack of a server report in our reports folder for our client. The lack of response via RWW or VPN made it clear we were in for an on-site visit.

Once on-site, the hard drive light was solid. There was no response from the server when a monitor, keyboard, and mouse were connected. This particular 1U box is headless.

We pressed the power button for about 5 seconds to power the server down and pressed it again to power it back up and we were greeted with:
LSI RAID: Cannot detect array configuration
The only option we had was to enter the RAID BIOS. The server would not boot up.

In the RAID BIOS, we needed to tell the RAID controller what was going on. We entered the View/Add menu, chose the Port 0 drive, and was able to see that the drive on Port 1 was failed. We then needed to save the status update to the RAID BIOS.

Since we were there first thing in the morning, we booted the server up and verified that SBS 2003 R2 Premium was fully operational. Everything booted up fine. So, we left the server in that state for the business day.

When we came back at the end of the business day, we shut down the server and replaced the defective drive. This is a non-hot swap unit, so there was a little work involved pulling things apart and putting them back together.

And now comes the really big caveat: The RAID controller on this particular motherboard requires the rebuild to be accomplished while in the RAID controller BIOS. Thus, we needed to be back again first thing in the morning to reboot the server into the SBS OS. This is really surprising since all of the Intel Desktop Boards we have used the on board RAID controllers would run the rebuild in the background.

The rebuild was successful, and we had a fully functioning SBS box when it booted up, but the rebuild situation for the array was an eye opener as far as the need to initiate it manually in the RAID BIOS. FYI: The rebuild rate for the 750GB RAID 1 mirror was about 4-5 hours while the server was offline.

A little more research is in order for this situation. The new S3000SH and S3210SHLX boards show a download for the Intel RAID Web Console 2 available on the product Web page. We need to see if we can use the RAID Web Console to initiate a rebuild from within the OS. Having the ability to initiate the rebuild while the OS is online can be critical to a business with only one server.

Yes, performance will definitely be an issue as the on board RAID controller is busy trying to rebuild the array onto the new drive. But, there are ways to work with that situation versus not having the server online at all.

We were not able to install an add-in RAID controller on this particular 1U box since the only available slot was taken by a PCI-E Gigabit NIC.

For our smaller clients, there can be a struggle between cost versus performance and added redundancy and hot swap features. Having those features is just like having an insurance policy.

In this case, the insurance policy of an add-in RAID card and hot swap would have enabled us to change out the defective drive and rebuild the array without downing the server. That insurance policy would have allowed the server to keep on functioning too. It has been our experience that a drive failure on an on board RAID array tends to lock up the server or workstation.

And, one other thing: With our client having their ShadowProtect backups in place, with the last incremental taken just before the hard drive locked up, helped us to be quite confident once the situation had been assessed. Worst case scenario we were replacing the drive, recreating the array, and running the ShadowProtect recovery to recreate the SBS box's partitions. We had all of the bases covered.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists

*All Mac on SBS posts are posted on our in-house iMac via the Safari Web browser.

Friday, 30 May 2008

Intel Server Board Memory Caveat

Each and every Intel Server Board, such as the S5000PSLSATAR, will have a list of tested memory for that board.

Tested memory list [PDF]

That list is far from static. It is constantly being updated as more and more manufacturers get their memory products validated with a particular server board.

Intel goes to great lengths for us in the Channel to make sure there is a list of components that will work as they should when configured in a server. These lengths give us the confidence to quote on a server setup knowing that we will have a stable server at the end of that particular build.

When we put a call into Intel to get some assistance with troubleshooting a server problem, one of the first questions out of the technician's mouth will be what type of memory was installed on the server board.

If that memory is not on the approved list, then the reseller/integrator will be responsible for purchasing and installing memory that is on the list before Intel will go any further with the troubleshooting.

This is especially important to keep in mind when faced with:

Kingston Product list for S5000PSL

Note that the part numbers with the "I" suffix are Intel validated. The others are not.

Given the plethora of memory chip manufacturers, PCB board manufacturers, and all of the other components on a memory stick, there are so many combinations that come together to defeat a server configuration out of the box or even a week, month, or year down the road.

The "I" suffix gives us the assurance that the components that make up the memory stick will work as expected when plugged into an Intel server board. The "I" memory sticks will cost more too. It is, however, a small price to pay when a server is due to be delivered tomorrow but decided to start spontaneously rebooting this afternoon near the end of its 72 hour burn in!

We were in that situation once ... and we learned a valuable lesson that day.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists

*All Mac on SBS posts are posted on our in-house iMac via the Safari Web browser.

Monday, 21 April 2008

SBS Setup and Intel (or other) Adapter Teaming

When the first phase of the SBS 2003 setup (blog post SBS setup checklist) is complete, one goes on to install all of the relevant drivers and configures the partitions for production service.

From there we move the swap file over to its dedicated partition and verify that the hardware and base OS are happy before proceeding on to second phase of the SBS setup.

One thing one should never do is team the LAN facing adapters before running SBS Setup Phase 2.

Remember that one of the questions we get asked once we have initiated the second phase is which adapter is the Internal one?

Since a teamed adapter is virtual in that it is comprised of two or more physical adapters, we run into a problem once SBS setup takes over: SBS setup only wants one physical adapter enabled during the process.

If SBS setup finds any other physical adapters enabled during phase 2, it disables them. The SBS setup routine does not figure out that when it disables the physical adapters, the virtual teamed one will disappear ... but that is exactly what happens.

The only time we will see any indication that there is a problem will be when we see the red X on Server Tools, and once SBS setup is finished with a list of errors including the inability to install SharePoint.

By then it is too late.

Timing wise and stability wise it is expedient to blow away the box and start fresh. Troubleshooting and trying to work around the errors presented is not advisable on a new SBS box.

This is one situation where if this mistake was made, we must choke on the cost of rebuilding that box ... and facing the music if we need to move the SBS install date at the client site.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists

*All Mac on SBS posts are posted on our in-house iMac via the Safari Web browser.

Thursday, 17 April 2008

Intel SRCSASRB RAID Controller and SR1560SF Caveat

Intel's SRCSASRB RAID controller has a catch when it is to be installed on the S5400SF series server board that comes in the SR1560SF series server system: The PBA on the card needs to be D92806-153.

The MM# on the box needs to be 892598.

The PCN:

Intel Product Change Notification: 108057-00

The above mentioned PCN came to us via the Intel support technician we spoke with on the phone today.

Now, we have mentioned in the past on a number of occasions how we need to research our system components to see if there were any compatibility issues relative to the revision level of the components.

It is very important, because in the case of Intel desktop and server boards one runs into the board revision level being too low for a particular processor on a somewhat frequent basis.

The same is true for compatibility between RAID controllers and a given desktop or server board.

In the Tested Hardware and OS List document we see the following:

Intel Server Board S5400SF Successfully Tested with SRCSASRB

Note the BIOS, BMC, FRU/SDR, and HSC revision levels required.

At the bottom of the Tested Hardware document we find:

SRCSASRB Stepping C1 not compatible with Gen2 PCI-E

Intel's products all of a PBA number on them. In this case, the product we pulled out of the SRCSASRB box had a PBA number of: D928906-152.

As indicated in the above PCN, we needed to have -153 in order to have the correct stepping for the second generation PCI-E slot that the SR1560SF/S5400SF has.

But, our box has an MM# 892598, which by the above PCN should have had a -153 in the box.

We did not however find out about the above PCN and the correct PBA number until after the SR1560SF would freeze during POST with the SRCSASRB installed and the Intel technician supplied us with a link to the PCN.

It is truly amazing how Murphy's Law finds its way into situations where timing is of the essence! ;)

Our supplier has shipped us another SRCSASRB to replace this one. Hopefully it is correctly boxed! 8*O

Product related links: Philip Elder
MPECS Inc.
Microsoft Small Business Specialists

*All Mac on SBS posts are posted on our in-house iMac via the Safari Web browser.

Thursday, 21 February 2008

SBS + ShadowProtect = Change system partition size on a RAID array

When we run into an SBS server that has a system partition that is getting full, we run into a bit of a bind if the subsequent partitions are in use with no free space to be had anywhere.

If we are working with a server that has one set of drives configured into a RAID 5 array with a system partition and a couple of additional partitions, what do we do?

With ShadowProtect, we have two rather simple options:
  1. Migrate the system and data partitions to a new array on larger drives
  2. Migrate the system and data partitions to resized system and data partitions - if there is free space available on the other partitions.
The first solution is pretty straight forward:
  1. Create ShadowProtect images of all partitions to USB HDD
  2. Power down the server
  3. Swap the old drives for new larger drives
  4. Configure the new RAID 5 array in the RAID controller's BIOS
  5. Create a bootable Logical Drive using all space on the new array
  6. Boot to the ShadowProtect Recovery Environment (SPRE) via CD
    • If necessary, load the appropriate RAID controller driver to access the array
  7. Create your new larger partitions via the Disk Manager in the SPRE
  8. Format them
  9. Restore your system and data partitions from the ShadowProtect images
  10. Verify drive letters on the new partitions
  11. Reboot
Once the server has finished booting, the new system partition size will be available for the additional SBS growth that is needed.

If budget does not allow for new larger hard drives, then we use the following method to reconfigure the RAID array's partitions if free space permits:
  1. Create a ShadowProtect image of the system partition to USB HDD
  2. Copy the other partition's data to the USB HDD in a respectively named folder
  3. Reboot the server
  4. Logon to the RAID controller's BIOS
  5. Delete the existing RAID array and its partitions
  6. Create a new RAID 5 array using the existing drives
  7. Create a bootable Logical Drive using all space on the new array
  8. Boot to the ShadowProtect Recovery Environment (SPRE) via CD
    • If necessary, load the appropriate RAID controller driver to access the array
  9. Create your system partition larger than it was before via the Disk Manager in the SPRE
  10. Create your subsequent partitions in the Disk Manager
    • Keep in mind the volume requirement for the data that will be copied back.
    • If there is not enough available space on the other partitions to give up for the system partition, then the above method with new drives is your only option.
  11. Format them
  12. Restore your system partition from the ShadowProtect image
  13. Copy the data back to the respective partition from its folder on the USB HDD
  14. Verify drive letters on the new partitions
  15. Reboot
You will now have a reconfigured partition setup.

Note that the GUID associated with the original partitions may no longer apply. Verify that any backup jobs (ShadowProtect uses the GUID to pick partitions for backup), application, or the OS itself has picked up the partitioning setup and correct drive letters.

You may need to reboot into Safe Mode and reassign the drive letters from within the OS for the additional partitions if they do not get assigned correctly by the OS upon boot up.

This methodology is relatively quick and quite painless.

The whole operation for Method 1 is no more than an afternoon depending on data volumes and server speed.

The operation for Method 2 would be a bit longer depending on how much physical data needs to be copied to the USB HDD and age of the server itself.

Philip Elder
MPECS Inc.
Microsoft Small Business Specialists

*All Mac on SBS posts are posted on our in-house iMac via the Safari Web browser.