Category: TechEd EMEA IT Pro

Speaker: Elden Christensen, Microsoft – owner of the failover clustering and network load balancing features.

One of the primary reasons that DR invocations plan fail is the dependence on people.  This was the result of a study after Hurricane Katrina in New Orleans.  In the event of a disaster people focused on their personal priorities, not on their DR plan actions.

Network Options To Stretch Cluster:

• Stretch the VLAN: 2003 did this
• Dissimilar subnets: 2008 introduced this support.

Longer distance = latency.  Windows 2008 allows you to tune the heartbeat time out.  Out of the box <500ms is fine but you can tune this.  This can be tunes differently for nodes on the same or different subnets within the one cluster.

Windows 2008 inter-node communication can be encrypted for cross WAN clusters.

Client reconnect reconsiderations:

• If the nodes are in different subnets then there are DNS timing issues to consider.  A records get cached on DNS and on the client.  If there’s a failover then what happens?  The client/DNS have cached the old record and clients fail to connect until their purge/timeout the DNS cache.  Also consider DNS AD replication between sites.
• Configure a smaller TTL for the record but you need to find the right balance between too frequent and infrequent lookups.
• RegisterAllProcidersIP and/or HostRecordTTL

Alternative 1: Advanced Planning:

Have a local failover in Site A and in Site B.  Configure the cluster to failover to a local node first, e.g. a local hardware issue rather than a site failure.  If site failure then fail over to site B.  This is OK if the DR plan allows for non-instant failover.

Alternative 2: Otherwise stretch the VLAN.

The IP of the clustered resource never changes.

Alternative 3: Abstraction Device

For example, Cisco has a device to abstract and IP address to reroute it as required to the correct server in the correct site.

Storage

You need to have two copies of the data.  Single site allows for single copy storage.  But that’s not going to fly for DR.  You need to replicate the data between site A and B.  MS relies on the vendors/partners, e.g. HP LeftHand, HP EVA Controller, HP XP Controller, Compellent, DoubleTake, SteelEye.  There is also application stuff such as Exchange CCR.

Synchronous or Asynchronous can both be used – it depends on your application.  Synchronous commits data to both sets of storage and then responds to the application to confirm the write.  Asynchronous writes to one set of storage and then replicates it to the other site.  Obviously the latter is good in limited bandwidth scenarios.  There is a potential for data loss.  It stretches over great distances and has no impact on application performance.

But the former guarantees no data loss but requires more bandwidth between sites.  Latency is an issue so the stretch is a short distance (<100KM) and has an impact on application performance with greater latencies.

The storage partner writes DLL’s that integrate into clustering so it ensures consistency of storage ownership/failover during a failover of the clustered resource.

The validation tool is not written for these replicated storage solutions and will fail.  This is acknowledged by MS and is documented online.

HP StorageWorks Representative

The speaker is talking about the HP story, CLX for Windows.  CLX = Cluster Extension Resource.  This is for EVA and XP SAN.  There is support now for Hyper-V Live Migration in the new release.  This adds W2008 R2 and Hyper-V Server 2008 R2 support.  This Live Migration support indicates the speed of failover.  EVA support in a month, XP next year.  Apparently this does not support CSV at the moment due to the controllers role in the replication process.  This costs around €3000 per cluster node so you better be serious about DR – and this doesn’t include SAN replication licensing.

We get a video of this demo based on W2008 R2 Hyper-V live migration on a pair of replicated EVA 4000 SAN’s.  We saw 3 failed pings on the grainy video but the HP guy claims they were retransmits, not dropped packets.  I’m not convinced that HP have real Live Migration between sites but 2-3 missed pings between sites for DR is pretty good.  You have duplicate copies of data in 2 sites in case of a disaster.

Quorum Overview

It’s all about getting a vote majority to decide who owns a resource.

• Disk only: The quorum disk (who ever is the owner of it) decides.  Even number of nodes.
• Node and disk majority: Disk owner breaks the vote
• Node majority: No witness disk
• File Share Witness: Instead of the disk

Replicated Witness Disk in DR:

Not to be used unless recommended by the storage vendor  Normally not used in replicated storage clustering because it is really 2 disks, one in each site.  MS not a fan of it.

Which to use?

• Node Majority: Odd numbers of hosts.  The majority of nodes will be in the primary data centre, e.g. 3 nodes in site A and 2 nodes in site B.  If there’s a break in comms between sites A and B then the nodes will vote.  If node 1 in site A (3 nodes) votes it can talk to itself (1 vote) and the other 2 (now 3 votes).  In site B node 4 can talk to itself and it’s neighbour (2 votes).  Therefore the resource stays in site A.  But in DR if site A burns down then you need to manually override because site B cannot win a vote.  This is called Forcing Quorum.
• Multi-Site with File Share Witness: This is normally the best one to use.  Place a file (SMB) share in a third (witness) site.  There’s nothing special about the share other than a single text file.  This allows even node numbers where the file share is the vote breaker.  If site A fails then site B can see the file share in the witness site.  Site B initiates a failover automatically.  But what it site A and site B can both see the witness site but not each other?  Seems there’s a solution with the file share but the speaker doesn’t say … I guess it’s something in that text file.  This comes up in QA.  The node that owns the file in the file share is healthy then it becomes the vote breaker.

Workloads

Hyper-V: If you use DHCP then you can use different VLAN’s.  If your VM’s use static IP then stretch the VLAN(s).  Live Migration really requires stretched VLAN’s because otherwise the IP must change in the VM and that requires a TCP outage.

CSV: Requires a single VLAN between nodes.  CAV assumes all nodes can concurrently access the LUN.  SAN replication assumes that only one array has the replicated LUN active at a time.  CSV is not a requirement for Live Migration.  MS says you should talk to your storage vendor for support statements.  The whole scenario depends on how the storage is replicated by the vendor.

SQL: Missed this because it was very quick.

Exchange 2007: It has CCR so you don’t need storage level replication.  Change the TTL to 5 minutes.  File share witness should be on the hub transport server in the primary site.  Exchange 2010 is probably very different because of the possibility of using a DAG.

Q&A

DFS-R: Can you use this for multi-site clustering? Yes and No.  DFS-R is supported on 2008 R2 clusters but you cannot use it as the replication mechanism because it only replicates at file level and file close.

Does the HP CLX support CSV? Not in this release.  They are working with MS to get this working.  HP LeftHand will do this.  Compellent does this too – I think Lakeland Dairies (Irish company) are using their solution for inter-building DR for Hyper-V on their “campus”.  I believe there’s a whitepaper
on it somewhere on the MS site.  I did find this video.

What is Springboard?  It’s a program by Microsoft to help IT Pro’s do a successful deployment of the desktop operating system, e.g. Windows 7.  The site acts as a portal, gathering together articles, blog posts and videos generated by experts inside and outside of Microsoft.  That means you can go to a single location to learn about :

• Windows 7 and its features
• Plan your deployment
• Run your pilot
• Deploy Windows 7 and
• Manage your desktop network

Springboard also runs the Springboard Technical Excellence Program (STEP).  This is a global group of around 140 MCT’s and MVP’s who work with Springboard to add content and present the content to local markets.  I’m a member of STEP.  I applied to staff the Springboard stand for a few hours a day at TechEd EMEA 2009 and was accepted.  That’s why I’m here in Berlin now.

Yesterday afternoon I did my first stint at the stand.  It was hectic.  Springboard sponsored a party for MVP’s, MCT’s and IT Pro’s and tickets were limited to 300 people.  The demand was nuts and not everyone managed to do their registration correctly.  Throw in the usual conference booth competitions and the genuine interest in Springboard and you can imagine how busy we were.  I ended up answering all sorts of questions:

• What is STEP?
• Can you tell me about Windows 7?
• Hyper-V
• Boot from VHD
• SAN Storage
• MDT 2010/WDS
• “I didn’t register for the party but can I get a ticket?”

It was wide and varied audience.  I’m back on again this afternoon.

We had the party last night at a club called Watergate.  I hung out with some of the Irish folks and didn’t go wild.  The drink was flowing and plenty of folks made the most of it.  I’m still feeling a little ropey this morning despite being cautious but I suspect it’s mostly lack of sleep over the last 3-4 nights that’s got me.  We have an Irish night out tonight and I’ll be taking it easy there too.

Speaker: Jeff Wettlaufer (Microsoft)

This rooms is packed.  Standing room only at this point. 

There will be support for 300,000 support.  Jan 10 2010, mainstream support for SMS 2003 ends.  Be aware of V.Next when planning your migration from SMS to ConfigMgr.

ConfigMgr 2007 SP2

SP2 adds support for Windows 7, Server 2008 R2, Vista SP2 and Server 2008 SP2.  This allows them to be clients and allows those servers to host site roles.  Server 2008 R2 BranchCache is a game changer.  In MS they cut 90% of traffic to the Mexico office with this.

Improved Client Policy Evaluation:

• Faster processions.  Pre-SP2 there was a 2 minute delay
• More efficient s/w deployment configured to run at user logon
• There was a 10 minute delay pre-SP2
• User/Group targeted advertisements instantly available after logon with SP2

BranchCache Support:

• Integration with ConfigMgr reduces vastly WAN traffic and transfer loads on distribution points.
• Client downloads are faster (from local cache)
• Configuration Document guide available
• Requires R2 DP and Win 7 Enterprise/Ultimate clients

64 bit Support

• App-V x64 client
• Remote Control for x64 XP and W2003
• x64 performance counters
• Update to OpsMgr MP for 64bit OS
• x64 support for OpsMgr agent

Asset Intelligence

• Software Assurance not required
• The requirement for a cert has been removed for updates synchronisation

Intel vPro Technology

• Out of band wifi management
• Power state configuration
• Other stuff – vPro costs extra on the h/w side

KVM coming in the future.

OS Deployment

• Multiselect and delete drivers in the console
• Task sequence editor UI displays the package names as in the console.

Available as of Oct 22.

Now on to ConfigMgr 2007 R3

It’s still quite early in the engineering phase.  Support for power management – it’s the only thing they’ve talked about so far from what I can see. 

• Monitor power consumption
• Plan and create a power management policy and check for exceptions
• Apply power management policy
• Check compliance and remediate
• Report in power savings and costs savings and environmental impact

Collections will be used to apply power policies so some engineering required to group clients appropriately.

Demo:

The Power Management Client Agent has a dependency on the Hardware Inventory Client Agent.  With both configured (and with schedule noted on hardware agent) you’ll start getting data back.

In the collection you can enable power management settings in a new Power Management tab.  There is an option for a peak plan and a non-peak plan.  That allows the admin to select the Windows power plans, e.g. balanced, etc.  In the properties you can configure that Windows power plan in the ConfigMgr console, e.g. sleep after 5 minutes, etc.  This will support older OS’s because the core focus is on power down and hibernation but Win7 takes advantage of it more.

I ended up coming in a bit late for this one due to meeting a few people I know.

So far what I’ve picked up is that DPM v3 can grow storage volumes as required.

DPM v3 can back up laptops over VPN.  This supports XP, Vista and Windows 7.  It scales to 1000 clients per DPM server.  Disconnected clients will continue to backup using VSS according to DPM policies.  This is not DirectAccess aware now.  It was a huge request but release timing prevented it.  I’d guess it’ll be in SP1.

What I missed:

• Self service restore for users from Windows Explorer (Previous Versions) or Microsoft Office.  Windows 7 clients get a system tray icon with a bit more information.
• Self service restore for SQL DBA’s
• The store scales much more than 2007
• We get R2 Hyper-V support CSV and alternative host recovery

Client policies can include specific folders and exclude specified file types.  Users can also include specific folders (optional for the admin via policy).  This is new for typical corporate backup.  Products live Iron Mountain’s Connected/LiveVault had this for ages.  You won’t get moaning alerts when the client is offline.  You get 14 days allowance for the client being offline – then you get an alert.  The client has a simple synchronise now button when they get back to the corporate network to synch their VSS backups to DPMv3.

Now they talk about DPM2DPM4DR, i.e. DPM replication to a DPM site in a DR site.  This is done by setting up another DPM site and then installing an agent on the primary DPM server.  You can do this in 2007.  A powershell cmdlet allows failover.  v3 includes a GUI click to do this.  You can do production site DPM to DR DPM to tape.  You can then use the DR tape to recover in the production site … or the DPM store.  The process of selecting what to backup from the primary DPM to the secondary is that you pick which items it has backed up.  So you can tier your data backup .. maybe not everything needs to be backed up to the secondary/DR DPM.  You also can decide to use a different synchronisation schedule depending on your bandwidth.  This is also a block level differential backup like the primary backup.

Changes:

• Scalability per DPMv3 server: 100 servers, 1000 laptops, 2000 databases
• Increased fan-in of data sources per DPM server
• Up to 80TB per DPM server
• Automatic protection of new SQL/MOSS data sources
• Decreased inconsistent replica errors
• Reduced alert volume
• Automatic rerunning of jobs and improved self healing

Worst case scenario is that you lose 14 minutes and 59 seconds worth of data because it backups up every 15 minutes.  By restoring SQL to “latest” it recovers not only the DB but replays the logs to the very last transaction that was committed in the TX log.

You can produce scheduled reports of backup status for protection groups, servers or data sources, e.g. a DBA can get reports on their databases in their mailbox every morning.

Bare metal recovery for Windows Servers is now a check box.

The 2007 – 2010 upgrade path is … You must be on 64 bit 2008 or higher OS to do an in-place upgrade.

Beta is out now.  The RC is after XMas 2009.  The RTM will be Spring 2010.

Non-AD machines will be supported in DPM 2010 (not in the beta.  It will be in the RC).  It sounds like it will be drowned in scripting.  I think they should talk to the OpsMgr team.  X.509 is the future.  OpsMgr isn’t as clean as it could be in this regard but it’s easy enough.

Speaker: Mark Minasi.

The session I was originally going to wasn’t what I was expecting so I decided to go to Mark’s VHD session instead.  As usual, these are just highlights from the session and you should attend it if you get the opportunity.

This session is all about VHD.  Mark warns us it’s nearly all command prompt.  It also applies to Windows Server 2008 R2.  As announced at TechEd Barcelona 2008 by Mark Russinovich, VHD is now the MS data centre image standard.

Survey: half of the audience have used WAIK and WinPE.

BCD/BCDEDit replace boot.ini since Vista.  That first 100MB volume in Windows 7 has no drive letter and is where the BCD is stored.  You only see in in Diskpart and Disk Management.  We also can use this for BitLocker instead of the nasty Vista (non) solution of a 1.5GB partition.  This also allows us to boot from VHD files.

Normally, on a clean build that 100MB partition is at the start of your drive.  However there are upgrade scenarios where it can appear elsewhere on the drive (end or middle).  It doesn’t matter and don’t bother to move it.

Example of CompletePC backup from command prompt:

wbdadmin start backup -backuptarget:m: –allcritical –incvlude:e:

The –allcritical flag includes the 100MB partition.  Do this.

CompletePC backup uses VHD as it’s destination instead of tape.  That first backup takes and age but all the folling ones are synthetic, i.e. differential with the effect of full.

Note: You can use Disk2vhd to convert a hard disk to VHD.  Not supported in production.

Notes:

• Server 2008 R2 uses VSS now so Server Backup is 5 times faster than Server 2008.
• DFS and FRS are gone from Windows Server 2008 R2.  Watch out for AF upgrades from 2003 R2 or earlier.
• You can do S/W RAID on Windows 7.  Don’t do it.

We can mount VHD’s and we can boot from them:

• Those product samples from MS supplied as VHD can be booted up on your lab machine, not only as a VM.
• You can multi purpose a physical machine, e.g. a server with multiple VHD’s can boot up as a web server 29 days a month and as a finance/tax machine on the last day of the month.  This is not dual booting.
• You can dual boot a demo laptop with Windows 7 for office work and Server 2008 R2 with Hyper-V for demo work.

VHD is 96% efficient, i.e. runs at 96% or thereabouts the speed of the underlying physical disk.

So you can have drives on your machine:

• 100MB BCD partition
• The rest of the disk: on here you place your VHD’s.

With BCDedit you configure BCD to boot from the VHD.  The VHD is mounted, boots up and becomes C:.

We will be using DISKPART – create vdisk to create the VHD.  These slides will fly past so I won’t blog them.

Here’s a note: MS got their terminology mixed up again.  The 3 types of VHD in DISKPART are:

• Fixed: all the VHD space assigned at once
• Expandable: This is known as dynamic to the rest of the world, including Hyper-V.  It only supplies disk space to the VHD as required.
• Differencing: It reads from a source but writes differences to the new differencing VHD

You can also do this in Disk Management GUI.  Note you cannot create a differencing disk in the GUI – differencing should only be used in labs.

You can use attach vdisk in Diskpart to mount/surface/attached the VHD.

Now you can initialise the disk and create a volume in it in Disk Management, etc.  It’s a new volume.  Disk Management is aware it’s a VHD.  Explorer will now present the new drive mounted (aka attached or surfaced).

Now you need to use BCDedit to configure a boot option for the VHD.

Now Mark does  a demo of a unsupported (he warns us) scenario on VMware Workstation (Virtual PC doesn’t cut it for demos).  He’s going to show how to do a empty C: drive that only contains VHD’s.  He runs Windows 7 setup.  Then SHIFT+F10.  He uses diskpart to create the 100MB partition and the rest of the disk as H:.  The 100MB is primary partition and active.  The rest of the disk is primary and marked as H:.  He exits diskpart and shows the empty H.  He creates a VHD folder in H: and goes back into diskpart to create his VHD (as above), selects it and attaches it.  He exits and returns to the Windows 7 setup.  The disk configuration now presents the H: and the VHD.  The VHD is not a possibility for installation  – but he can do it anyway!  It is not supported but it’s a great way to set up a lab machine.  You will lose hibernate with this setup.

An MS executive I’ve never heard of (Stephen Elop) is speaking.  2 things are keeping me from attending.

1. Last year’s keynote was dreadful.
2. I walked down that way and it reminded me of a cattle crush at a slaughterhouse.  5000+ people (there are 7500 in the venue this week) were trying to squeeze through 1 door.

I’ve managed to get access to one of the way too rare power sockets so I’m doing some work.

—-

*This is posted after the event ended*

Screw it – I came down 10 minutes after the session started.  I expected to not find a seat.  Boy, was I wrong.  People were leaving the room in their droves.  There was a conversation going on between a bunch of executives on the stage that was BORING!  It was out of touch with the technical audience and felt very scripted and rehearsed.  I walked in and got a seat within 10 seconds.  My biggest problem was stepping by the crowds leaving.

As I type this I can’t hear the stage over the sound of people leaving.  I am not exaggerating.  This session is worse than the Visual Studio one at TechEd in 2005.  Whoever planned this one really didn’t think of the non-executives.

I wonder why there’s no wifi here.  I can’t tweet anything from the venue.  Plenty of peer-to-peer wifi honeypots in this room though 🙂  It’s always the same at TechEd.

Stephen Elop starts talking about Windows 7.  I don’t know how bad he thinks MS subsidiaries have been but he’s telling us stuff we’ve known for a year at this point.  I guess I’m not alone because people are still leaving by huge amounts.  Last time I saw this sort of departure was at a DSI talk by an MS executive in the USA who couldn’t speak English.  This talk is going down like a dead balloon.  It’s high level marketing speak for the wrong audience.  I wonder if people can see this exodus on the live feed?

I also suspect that having the keynote late in the day on the 20th anniversary of the “fall of the wall” was a very bad idea.  Sure, it gets the USA online audience but at the cost of the local audience.  Every session I’ve gone to so far has been full or near full with high attendee retention.  This keynote has played a bum note for a lot of people.

Exchange 2010 Launch

Now we get the Exchange 2010 demo.  They’ve done a demo of a live mailbox migration in front of “7000” (minus the 1000 that has left so far) people.  Conversation threads, mobile, OWA all demonstrated.  UC presence awareness is shown in OWA.  Transport protection rules are shown.  Elop is using Firefox for his OWA part of the demo.  The local audience applauds with laughter. 

You know what?  I think Patrick in MS Ireland got a lot more done on Exchange 2010 and Outlook 2010 in his half hour during the community launch tour than these folks did.

I’m not on the right drugs.  There’s a video of a dude playing with a man in a Fox suit.  What the hell is that supposed to be?

ForeFront

Exchange 2010 is not the only launch today.  ForeFront is being launched as well – that’s logical.  And that’s all he has to say on that!  I feel for any of the product teams who are here.  They will have worked hard for 2-3 years to get a one liner for their launch.

Windows Server

The launch was done on October 22nd.  72% of global servers run Windows Server of some sort.  Hyper-V and Live Migration get a plug, along with System Center.  Out they roll Continental Airlines once again.

People still leaving.  I think they should have had 3 launch events: IT Pro, Executive and Developer.  The 3 audiences have different needs.

Robert Wahbe (corporate VP on Server) now talks.  To be honest, most of us will probably have heard most of this already.

Some claims about time savings and power savings are made.  Naturally, these are all company dependent.  For example.  If a physical machine consumes 30% of a Quad Xeon on a dedicated physical machine then it will consume slightly more than 15% of CPU on a dual quad core Hyper-V host.  There are no magic savings beyond what your company needs.

Jeff Wettlaufer is up now after the marketing speak. 

…

I’ve had enough.  I’m outta here.  I met up with Alex Yushchenko outside.  It was agreed; the keynote was dreadful.  The flow of people leaving just kept growing and growing.  I know Robert Wahbe did a double take at one point when he saw the queue of people trying to get out the single doorway.

The keynote was inappropriate for this audience.  They’re techies who don’t respond to marketing presentations.  50% of them were devs who got nothing from the session.  In fact, the European devs miss out because all the Azure stuff (and probably VS 2010) happens at PDC next week in the USA.  The IT Pro’s were bored.  Executives don’t come to TechEd.  Timing wise, TechEd Europe is late in the year.  All the big stuff is done by now, e.g. Win7 and Server 2008 R2 announcements were in 2008 in Barcelona and TechEd USA in the Spring.  MS needs to think again before repeating this mistake.

I still look forward to the technical sessions during the rest of the week.  It’s just a pity the keynote couldn’t have been a better use of the 2-3 hours.

Speaker: Rhonda J. Layfield (deployment MVP, speaker, author, trainer and journalist)

Subject: WDS.

Quick survey: More people are using WDS than have deployed Vista at work.  Not many deployed Vista out of the full room.

Rhonda says she is writing a book on her own: WDS, WAIK, MDT, ACT and volume activation.  I knew she had plans.  Great to hear this is on the way; it’ll be a good read.  Rhonda’s chapters in the Mastering Windows Server books in the past have been excellent and practical.

There are quick start guides for WDS on the way.

LOL – Rhonda not touching WSIM because “it’s a half day issue”.  Very understandable because that’s when all the questions will fire up.

WDS requirements:

• Active Directory
• DHCP (make sure DHCP server and WDS are both authorised in the DHCP console by an enterprise admin).  If WDS server is also DHCP server then you are prompted to create two 2 DHCP options.  Be aware of this because you will need to do this manually if you forget.
• DNS
• NTFS partition for the images

It is strongly recommended that you do not store your WDS images on the same volume as the operating system.

When you are setting up a pre-staged computer object in AD then pad the start of the GUID/MAC with 0’s until you have 32 characters.  You cannot continue with the dialog until you have exactly 32 characters.  Ouch – demo Gods were unkind to Rhonda and her demo of authorising new machines didn’t work out first time.  The WDS Server(s) must have permission to create computer account objects in the relevant OU/container.  Admin must have r/w permissions on the RemoteInstallMGMT folder. 

Delegation of domain join can be done with a registry edit on the WDS server. However, non-English DC’s can cause problems because group names might include non basic characters, e.g. á,ó, etc.  Use just a-z, 0-9, etc.

WDSUtil is the tool you’ll use to add a VHD to a WDS server.  It takes a while to add the image.

Permissions for the Service Control Point must be correct for WDS to work.  Use ADSIEDIT.MSC.  OU=Domain Controllers -> CN = WDSSERVER –> CN=WDSServer-Remote-Installation-Services Properties.

Network Issues

W2008 WDS increased block size from 512 bytes to 1,456 bytes.  If your network has TFTP block size of less than 1,456 bytes then this breaks WDS.  Install KB975710 + add REG_DWOWR MazimumBlockSize (value 512-1456 based on what your network can handle) to registry (missed the key – go search for it).

If WDS is on a different subnet to the client then you can configure option 66 and 67 to tell the client where the WDS server is.

If WDS is also DHCP then configure option 60.  Set up “do not listen to option 76”.

General

Renaming WDS: WDSUtil to uninitialize and initialize it.

ImageX doesn’t require a Sysprep.  Of course, WDS capture does.  If you forget then the volume to capture dialog will be empty.  Sysprep generalize Vista and Windows 7.

LOL – Rhonda got confused by WDS capture like I did.  Even if you choose to store the image on the WDS server you must create the image on a local volume – this requires sufficient space.  It also takes much longer than an ImageX capture because it does some verification (cannot be turned off).

Use image groups: create specific ones for OS versions and architectures, e.g. Windows 7 x86, Windows 7 x64, Vista x86, etc.  This allows single instance storage between WIM files in the image group to save significant disk space.

Tip for WinPE: Don’t create your own.  Use the ones from boot.wim from the sources folder in Windows 7 or Windows Server 2008 R2 DVD.  Try to always use the latest on because it has support for all of the latest operating systems.  Using a legacy one for a new OS gives you an “Operation failed with 0x80070002 The system cannot find the file specified” error.  This event applies to SP1 versions!

WDS will try to use IGMP v3 and then fail back to V2 when doing multicast.  Work with the network team to configure routers and switches.

Remember that PXE is broadcast based.  So DHCP needs to be forwarded on the network in multiple VLAN networks.

I’m going to be blogging each session that I can attend like I did last year.

It’s still Sunday and I’ve wandered into the not-yet-open exhibition hall.  I’ve a briefing in here in 45 minutes so it’s fine – I guess!

It’s quite flash looking in here.  You must walk pas the “The NEW Efficiency Zone”.  That appears to features lots of tech like Windows 7, Server 2008 R2 and Exchange 2010.  It is surrounded by System Center booths.  I can’t call them booths because they look way too flash.  There’s lots of “chrome” on gloss black.  Further into the hall are the more typical ask-the-experts stands.  If there are 3rd party stands here then they are way at the back.

The motto for the event appears to be “Efficiency”.  There’s lots of banners as you go up the entrance escalators featuring the above products and HP.  Strange how it isn’t IBM, eh?  LOL!  Who’d associate efficiency with IBM!!!

Exchange 2010 and the Office products obviously are playing a big role this week.  Exchange 2010 is being launched tomorrow in the keynote which is actually at the end of the day for a change.  There’s also a big push on SQL 2008 R2.  There’s a 2.5 hour session every day on installing it.  I guess it must be pre-booked or something but I’ll miss it due to my work schedule anyway.

I plan on going to Rhonda Layfield’s morning session on deployment troubleshooting and playing it by ear.  I’ll be going to some of the virtualisation stuff to see what’s happening.  It’s rare (even as a virtualisation MVP) that we get to hear directly from the MS team members.  There’s also some sessions on new features of Configuration Manager 2007 SP2 and 2007 R2, not to mention some V.Next stuff.  I was privvy to some of that info as a ConfigMgr MVP but it’s been a year since I could find time to attend their monthly briefings.  So I want to catch up.  I’ll obviously be dropping into one or two of Mark Minasi’s sessions to say “hi” to him.  Mark is doing 5 sessions in 2 days before flying to WinConnections in Las Vegas on Wednesday so he’ll be a busy man trying to prepare.

I thought strongly about signing up for Speaker Idol this year.  I had a topic that I though was cool.  But time was my enemy as usual.  Things just fell into place too late.  Anyway, that gives me a year to prepare for next year.  I have my demo environment (demo in a 3 minute presentation, YES!) and my topic.  TechEd 2010, watch out!

So I am waiting outside theatre 2 in TLC (The Learning Centre – maybe the exhibitors are elsewhere) for my briefing in 30 minutes.  There’s one or two others hanging about.  There’s no wifi in this room so I’ll be posting this later when I get a chance.

—

Added in the hotel 3 hours later …. We had the briefing where we were reminded that we are all effectively the face of MS this week.  We all got MS staff office shirts (in royal blue).  We were told about the layout of the venue.  The “learning centre” with all the MS stuff is in a different hall to the 3rd party exhibitors.  In fact, you have to pass them and the open comms room to even get to the very distant conference rooms.  I hung around afterwards and chatted to Irish fellow MVP (Clustering), Edwin Van Mierlo.  It’s an early night for me tonight.  I’ve got CNN International and BBC World News to put me asleep.  More tomorrow.

I’ve just registered at the Microsoft TechEd EMEA 2009 Berlin venue and I’m typing this while online on the event commnet.  I’ll be staffing the MS Springboard stand in the afternoons all week unless there’s a schedule change.  In the mornings I hope to attend as many sessions as possible.  Like last year, I’ll be blogging while listening.  And like last year, if you want, you can follow my postings on this link.

There’s some stuff open.  The hands-on-labs appear to be open as does the exam centre where you can do an exam at 25% off.  There an abundance of coffee and coke as usual.  I must remember to get some coke for the hotel room.  Luckily I remembered to bring my European power converters – unlike last year when Nathan Winters bailed me out.  I even brought a 4-way power strip to keep the laptops running and the phones charged.

I accidentally met up with a bunch of the MS Ireland folks in Dublin airport.  It turns out there aren’t too many flights from Dublin to Berlin.  We split a taxi from the airport which kept the costs down.  There hotels are some of the official ones.  They’re 4KM’s from the venue in the centre of the city.  I was cheap.  I went off the official listings and got a place about 15 minutes walk from the venue.  For 6 nights it’ll cost me 1/3 of the official hotels.  I am paying for it myself so I want to stay economic.  It turns out to be not a bad place.  It’s clean and close to here.  It’s cheap and has wifi.  I’m here to work and learn instead of partying so it suits me.

I’ve a meeting in 45 minutes where I’m to be briefed on working at a MS stand at a conference.  I guess I just need to be my usual outspoken self and to hell with whoever is in earshot 😉  I must save my line about successful map making and successful war making!!!  Is that Basil Faulty screaming?

Anywho, I’m gonna see what’s in the fridges.  I’ll also be tweeting a bit on twitter if you want to keep up.