Hyper-V Issues<br />Things I’ve Encountered<br />Aidan Finn, MVP - MicroWarehouse<br />
About Aidan Finn<br />Technical Sales Lead at MicroWarehouse<br />Working in IT since 1996<br />MCSE & MVP (Virtual Machin...
Agenda<br />Don’t know what new info you’ll get from this<br />But at least you’ll find out what issues I’m seeing and rea...
Assessment<br />“Measure twice – cut once”<br />How can you do virtualisation without knowing what’s required?<br />Gut fe...
Design Supervision<br />... or lack there of.<br />Typical scenario<br />Customer divides up the virtualisation project to...
Persistent Reservations<br />Storage goes offline<br />Number required = Hosts * CSV * Storage Channels/Host<br />Check wi...
Storage Offline & Host 9e BSOD<br />Check times of BSOD VS backup schedules<br />If it happens at same time as CSV backup:...
Third Party Backup & Replication<br />Watch out for 3rd party software storage with DR replication feature<br />CSV backup...
Storage is Slow - Backup<br />Storage is unexpectedly slow – Redirected Mode<br />Check the CSV backup strategy<br />Does ...
Storage is Slow - RAID<br />Am seeing people go budget on their SAN disk to save money<br />Slower disk at RAID5 for all C...
Storage is Slow - VHD<br />Some still advocating that Dynamic VHD is nearly as fast as Fixed VHD<br />True in the perfect,...
Antivirus<br />People are not following the guidance: http://support.microsoft.com/kb/961804<br />They scan CSV, VHDs, con...
Cluster Networking<br />I’ve seen companies:<br />Following W2003 or SQL 2008 cluster guidance<br />Wasting money on an ex...
Multi-Site Clusters That Aren’t<br />Scenario<br />Company has two offices near each other<br />One will be DR for the oth...
Lack of Patching<br />Incredible number of installs with no patching & Hyper-V is blamed:<br />iSCSI memory leaks (pre-SP1...
SBS as a Guest<br />Increasingly common<br />Seeing a growing trend with networking failures<br />The usual suspect (KB974...
Linux VMs<br />Dynamic MAC address leading to lost network access after migration<br />Are integration components being ke...
Snapshots<br />Most products that matter don’t support them:<br />AD, SQL, Exchange<br />Beware unmerged snapshots:<br />N...
NIC Teaming & Network Security<br />We know the official line on support<br />Beware NIC teaming features and VLANs being ...
System Center as a VM<br />Fine in theory<br />However:<br />Something should not monitor itself<br />Have seen SCVMM and ...
Windows Server VM Licensing<br />HUGELY common problem on clusters<br />Typical after P2V or on VMware sites<br />P2V’d OE...
Dynamic Memory<br />.BIN file matches physical RAM allocation<br />Is there enough room on disk to grow?<br />People getti...
Snapshots<br />Maybe supported by Hyper-V PG but not supported by AD, SQL, Exchange<br />Required shutdown/merge not obvio...
Thank You!<br />Aidan Finn<br />MicroWarehouse<br />Email - AidanFinn@mhw.ie<br />Web - http://www.mwh.ie<br />Personal<br...
Upcoming SlideShare
Loading in …5
×

Top Hyper-V Implementation Issues

9,822 views

Published on

This presentation describes many of the mistakes I've encountered/heard about with Hyper-V implmentations.

Published in: Technology
0 Comments
4 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
9,822
On SlideShare
0
From Embeds
0
Number of Embeds
1,351
Actions
Shares
0
Downloads
0
Comments
0
Likes
4
Embeds 0
No embeds

No notes for slide

Top Hyper-V Implementation Issues

  1. 1. Hyper-V Issues<br />Things I’ve Encountered<br />Aidan Finn, MVP - MicroWarehouse<br />
  2. 2. About Aidan Finn<br />Technical Sales Lead at MicroWarehouse<br />Working in IT since 1996<br />MCSE & MVP (Virtual Machine)<br />Experienced with Windows Server/Desktop, System Center, virtualisation, and IT infrastructure.<br />Blog: http://www.aidanfinn.com<br />Twitter: @joe_elway<br />
  3. 3. Agenda<br />Don’t know what new info you’ll get from this<br />But at least you’ll find out what issues I’m seeing and reading about<br />A lot of implementation issues are due to lack of education or documentation<br />
  4. 4. Assessment<br />“Measure twice – cut once”<br />How can you do virtualisation without knowing what’s required?<br />Gut feeling is insufficient<br />MAP is a starting point<br />I keep encountering people who don’t do assessments<br />And strangely they have issues later on!<br />Indicator that there will be later implementation issues<br />Assess for as long as possible to size accurately.<br />
  5. 5. Design Supervision<br />... or lack there of.<br />Typical scenario<br />Customer divides up the virtualisation project to many service providers<br />Servers, storage, network, Hyper-V, VMM, OpsMgr, backup, etc<br />Service providers can/will not cooperate<br />No one has design oversight<br />Things fall apart<br />
  6. 6. Persistent Reservations<br />Storage goes offline<br />Number required = Hosts * CSV * Storage Channels/Host<br />Check with storage expert<br />Beware systems like HP P4000<br />Hosts have 2 channels to every node in storage cluster<br />Solutions:<br />Is the storage firmware up to date?<br />Check storage design – all those CSVs required?<br />
  7. 7. Storage Offline & Host 9e BSOD<br />Check times of BSOD VS backup schedules<br />If it happens at same time as CSV backup:<br />Check the VSS provider<br />If it is Hardware VSS provider:<br />Check for latest version<br />Check for vendor support of CSV backup<br />Even with support, can be flaky H/W VSS provider<br />May have to switch to:<br />System VSS provider<br />Serialized backup<br />
  8. 8. Third Party Backup & Replication<br />Watch out for 3rd party software storage with DR replication feature<br />CSV backup will create snapshot on the replicated volume<br />Will cause replication/bandwidth issues<br />Encountered 3rd party backup with “2008 R2 Hyper-V support”<br />Had no concept of cluster & VM placement awareness<br />
  9. 9. Storage is Slow - Backup<br />Storage is unexpectedly slow – Redirected Mode<br />Check the CSV backup strategy<br />Does it really need to be hourly?<br />Are VMs with common backup strategy on the same CSV?<br />Are VM VHDs placed on many CSVs?<br />Strategy<br />1 CSV : 1 backup policy<br />Infrequent CSV backup (nightly/weekly/monthly)<br />Frequent in-VM data backup (hourly, half day, etc)<br />Remember: the entire CSV goes into redirected mode<br />
  10. 10. Storage is Slow - RAID<br />Am seeing people go budget on their SAN disk to save money<br />Slower disk at RAID5 for all CSVs<br />They find VM storage is significantly slower than pre-P2V physical server storage<br />Complicated with advanced storage concepts like disk groups<br />Implementers failing to grasp that virtual requirements are the same as physical requirements<br />
  11. 11. Storage is Slow - VHD<br />Some still advocating that Dynamic VHD is nearly as fast as Fixed VHD<br />True in the perfect, small, short-lived lab<br />Not true in the real world:<br />Fragmentation of dynamic VHD<br />Have been told that some storage controllers don’t deal well with random nature of fragmented storage<br />Rapid data growth leads to storage latency<br />Dynamic VHD on CSV can cause redirected I/O to grow if VM not on the CSV coordinator<br />
  12. 12.
  13. 13.
  14. 14. Antivirus<br />People are not following the guidance: http://support.microsoft.com/kb/961804<br />They scan CSV, VHDs, config files and processes<br />Lack of awareness<br />The security officer told them to “or else”<br />VMs are corrupted or disappear<br />0x800704C8, 0x80070037 or 0x800703E3<br />I hate AV on Hyper-V hosts<br />System, manual, or update errors<br />
  15. 15. Cluster Networking<br />I’ve seen companies:<br />Following W2003 or SQL 2008 cluster guidance<br />Wasting money on an extra “cluster communications” network<br />You really need:<br />Parent<br />VM<br />CSV / Cluster Communications<br />Live Migration *<br />Storage 1 & Storage 2<br />Maybe a backup network<br />Cable/enable network connection one by one<br />Label each network connection according to role<br />
  16. 16. Multi-Site Clusters That Aren’t<br />Scenario<br />Company has two offices near each other<br />One will be DR for the other<br />“Fast” 10MB+ link<br />They tell the implementer that it is a single site<br />Hyper-V and storage clusters are implemented as a single site cluster – but should be multi-site<br />Split brain scenario when that link eventually fails<br />Follow best practices: e.g. File share witness in 3rd site<br />Active-active sites & backup: VMs & CSVs<br />Redirected I/O across WAN link!<br />
  17. 17. Lack of Patching<br />Incredible number of installs with no patching & Hyper-V is blamed:<br />iSCSI memory leaks (pre-SP1)<br />Intel Nehalem/Westmere 1a BSODs (pre-SP1)<br />Still have patching to do since SP1<br />http://social.technet.microsoft.com/wiki/contents/articles/3150.aspx<br />Clustering for W2008 R2 SP1:<br />http://social.technet.microsoft.com/wiki/contents/articles/list-of-cluster-hotfixes-for-windows-server-2008-r2.aspx<br />
  18. 18. SBS as a Guest<br />Increasingly common<br />Seeing a growing trend with networking failures<br />The usual suspect (KB974909) is not the solution<br />Fix: Unknown to me!<br />Discussed with Microsoft PFE’s: disable advanced NIC features like TOE in the host and retry<br />
  19. 19. Linux VMs<br />Dynamic MAC address leading to lost network access after migration<br />Are integration components being kept up to date?<br />Integration components not updated automatically by VMM<br />Not quite as easy to do as with Windows guests<br />No VSS so needs specialised backup strategy<br />And consideration when placing on CSV<br />
  20. 20. Snapshots<br />Most products that matter don’t support them:<br />AD, SQL, Exchange<br />Beware unmerged snapshots:<br />Not immediately obvious in the GUI<br />Over time: fills disk, slows storage, causes app weirdness<br />People doing silly things:<br />Deleting AVD<br />Changing VHD<br />
  21. 21. NIC Teaming & Network Security<br />We know the official line on support<br />Beware NIC teaming features and VLANs being used for network security<br />HP NCU & promiscuous mode:<br />Page 24 on http://h20000.www2.hp.com/bc/docs/support/SupportManual/c02784628/c02784628.pdf<br />Recommends NCU vNIC and Hyper-V vSwitch for each VLAN for network security<br />
  22. 22. System Center as a VM<br />Fine in theory<br />However:<br />Something should not monitor itself<br />Have seen SCVMM and OpsMgr as VMs on production Hyper-V cluster<br />How does this do PRO/alert you if the host they are on has networking issue?<br />Maybe dedicated host/cluster for management VMs<br />
  23. 23. Windows Server VM Licensing<br />HUGELY common problem on clusters<br />Typical after P2V or on VMware sites<br />P2V’d OEM<br />OEM tied to original physical server<br />Licensing VMs with individual purchases of Standard edition<br />Allowed to migrate once every 90 days<br />License 2 host cluster, 8 VMs, with 2 * Enterprise<br />Not legal when 5+ VMs on one host (failover)<br />
  24. 24. Dynamic Memory<br />.BIN file matches physical RAM allocation<br />Is there enough room on disk to grow?<br />People getting cute with applications that have configurable memory caching?<br />Let apps work as normal<br />SQL Server<br />Check for edition support (Enterprise +)<br />Set VM memory buffer to 5%<br />NUMA – Is performance hit caused by NUMA spanning bad enough to disable NUMA spanning?<br />Memory leaking apps will love Dynamic Memory<br />Default maximum = 64 GB RAM<br />
  25. 25. Snapshots<br />Maybe supported by Hyper-V PG but not supported by AD, SQL, Exchange<br />Required shutdown/merge not obvious in GUI<br />People finding all sorts of ways to ruin VMs, e.g. delete a VHD<br />
  26. 26. Thank You!<br />Aidan Finn<br />MicroWarehouse<br />Email - AidanFinn@mhw.ie<br />Web - http://www.mwh.ie<br />Personal<br />Twitter - @joe_elway<br />Blog – http://www.aidanfinn.com<br />

×