IBM Cloud is upgrading a datacenter and users will have to halt their Power VMs
- Reference: 1706165051
- News link: https://www.theregister.co.uk/2024/01/25/ibm_cloud_toronto_migration/
- Source link:
For starters, tenants need to log a ticket with the subject "TOR01 Customer Data Migration." That ticket should include "at least two preferred dates between 2/12/2024 and 3/1/2024 and times (Eastern Time) for moving your existing active virtual machines and data to the new TOR01 infrastructure." Customers will also need to advise IBM of their Project ID, VM IDs, and the OS installed on each VM.
The deadline for filing those tickets was January 20. If you forgot, IBM will assign you a time.
[1]
Tenants are advised to "Reserve five hours per time slot (3AM to 8PM Eastern)," during which "Up to two VMs will be migrated." We shudder to think what this means for folks with big VM fleets in TOR01.
[2]
[3]
VMs must be stopped before migration – making this a disruptive event. "Operations will notify you when the migration completes, and you can restart your VMs," states an IBM FAQ, adding "After you restart your VM, you will be up and running in the new infrastructure, and your migration will be complete."
Err … not quite.
[4]Ex-IBM staff ask US Supremes for help in bringing age-discrimination battle to court
[5]IBM takes a crack at 'utility scale' quantum processing with Heron processor
[6]IBM Cloud to 'uplift' prices by up to 29 percent
[7]IBM sells off cloud business – yes, we mean Weather.com
On March 15, IBM will replace networking equipment in TOR01. Tenants will need to have configured redundant network links, or their VMs could lose connectivity for up to six hours during the upgrade.
Clouds are mostly sold as offering extraordinary resilience. Hyperscale operators shift workloads to new hardware without disruption – perhaps even without users being aware of it, unless they’ve specified bare metal resources.
[8]
IBM's Power – which Big Blue insists is terrifically resilient and reliable compared to the upstart world of x86 – clearly entails a different experience. At least in Toronto. ®
Get our [9]Tech Resources
[1] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offprem/paasiaas&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2ZbI-XV7pPAZMXQUlFYWAVwAAAco&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offprem/paasiaas&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZbI-XV7pPAZMXQUlFYWAVwAAAco&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offprem/paasiaas&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZbI-XV7pPAZMXQUlFYWAVwAAAco&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[4] https://www.theregister.com/2024/01/23/ibm_supreme_court/
[5] https://www.theregister.com/2023/12/05/ibm_heron_quantum_processor/
[6] https://www.theregister.com/2023/09/05/ibm_cloud_price_uplift/
[7] https://www.theregister.com/2023/08/22/ibm_weather_company/
[8] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offprem/paasiaas&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZbI-XV7pPAZMXQUlFYWAVwAAAco&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[9] https://whitepapers.theregister.com/
I have visions of a man with a brown lab coat and a trolley wheeling decks of cards to their new operator pigeon holes.
They have - LPM. For.. .ages. Most probably fired everybody who can use it, though.
IBM Live Partition Mobility
But Live Partition Mobility has some caveats with regard to connectivity and management. The source and target systems have to be managed by the same HMC(s) (which also puts version compatibility in the frame), and there has to be a quite significant amount of direct nettwork connectivity between the two systems. In addition, both systems need to be able to see and use the same backend SAN storage, if that is used, and it helps if it is IBM SVC or a Storwize or FlashSystem product..
I suspect that there is some storage migration going on here as well. If it was just a case of powering down the Power VMs, swinging storage across to another Power system by changing the mapping, and bringing it back up, manipulating the boot volumes as you go. It would take minutes, not hours per VM (I've done a medium-scale migration of a few dozen AIX LPARs like this in the past, and it works quite well).
All this assumes that the Power systems are runnning PowerVM. The OpenPower systems don't (but I believe can take part in other, mainly Intel based VM hurders), PowerVM is mentioned in the title, but not in the body of the article, and I failed to see any references to follow.
I am facing a Power8 to Power10 migration again in the near future, and I'm horribly afraid that the management will try to get IBM Consulting in to try to do a LPM, but the preparation work to do this would be horrendous! There would be more downtime bringing the firmware, OS and management infrastructure up-to-date than there would be just swinging the storage over and re-booting.
IBM Power systems do have these things available, but there is setup that needs to be done in order to use it, which is best done when the systems are first installed, They also rely on OS, firmware and management software being kept up-to-date.
Re: IBM Live Partition Mobility
>The source and target systems have to be managed by the same HMC(s)
not correct for several years now.
> They also rely on OS, firmware and management software being kept up-to-date.
As with most jobs.
Meh, At least...
...users aren't voting against Democracy for "reasons."
IBM don't have the cloud equivalent of VMotion or VEEAM replication ?
This is extremely surprising.....and smells a wee bit off.