|
View previous topic :: View next topic
|
| Author |
Message |
martin schlatter
New User
Joined: 06 Jun 2013
Posts: 9
Location: Germany
|
|
|
|
| How are mainframe updates installed and tested without a longer downtime? |
|
| Back to top |
|
 |
Robert Sample
Global Moderator
Joined: 06 Jun 2008
Posts: 8696
Location: Dubuque, Iowa, USA
|
|
|
|
| IBM provides SMP/E (System Modification Program/Extended) which is a program that installs operating system updates. Most sites have a "sandbox" LPAR which is used by the system programmers to install and test operating system updates. This testing can be extensive and last for weeks (or months) to ensure as few problems as possible when the updates are migrated into production. The downtime is minimal because all the testing has already been done so the upgrade process may be as little as changing the disk volume used to start the system (1 character) and starting it up. |
|
| Back to top |
|
|
martin schlatter
New User
Joined: 06 Jun 2013
Posts: 9
Location: Germany
|
|
|
|
| So you have a copy of the system on a second set of disks and there you install the new version. Then you test and the next time you boot from the new set of disks? |
|
| Back to top |
|
|
Robert Sample
Global Moderator
Joined: 06 Jun 2008
Posts: 8696
Location: Dubuque, Iowa, USA
|
|
|
|
| In our case, we actually have 3 sets of disks and use a grandfather / father / son rotation sequence so the grandfather set is the current production system, the father set is the sandbox set being updated, and the son set is the previous production system (to allow reversion should some serious production problem be uncovered). When we IPL (which is the mainframe term for a boot), we have to provide the address of the disk to load the system from. This value is changed to reflect which set is being used (for production as well as the sandbox system). |
|
| Back to top |
|
|
Willy Jensen
Active Member
Joined: 01 Sep 2015
Posts: 712
Location: Denmark
|
|
|
|
| Bigger sites use a sysplex where system software is shared between a number of systems. An less import member of the sysplex is started (iplled with the new level of software, typically like Robert Sample outlines, and only if no problems are encountered then the other members of the syspex are started with the new level. Applications are duplicated across the sysplex so that users should experience zero downtime. |
|
| Back to top |
|
|
vasanthz
Global Moderator
Joined: 28 Aug 2007
Posts: 1742
Location: Tirupur, India
|
|
|
|
| Quote: |
| In our case, we actually have 3 sets of disks and use a grandfather / father / son rotation sequence so the grandfather set is the current production system, the father set is the sandbox set being updated, and the son set is the previous production system (to allow reversion should some serious production problem be uncovered). When we IPL (which is the mainframe term for a boot), we have to provide the address of the disk to load the system from. This value is changed to reflect which set is being used (for production as well as the sandbox system). |
That's how we do it as well |
|
| Back to top |
|
|
|
|
