Product Documentation
Abstract
There is a performance problem when upgrading Information Server 11.7.0.1 with Information Server Enterprise Search installed to version 11.7.1.0. This document describes recommended steps which should be taken before the in-place upgrade to version 11.7.1.0; additional steps are needed after completing the in-place upgrade.
Content
The in-place upgrade of Information Server Enterprise Search (ISES) is a time-consuming process when upgrading from 11.7 Fix Pack 1.
The expected upgrade time depends on the Enterprise Search data volume. Typically the upgrade speed is approximately 20000 assets per hour, but the speed might be even smaller when processing over a million assets. Additional time is also required to upgrade an environment, even with no data.
Hence, the recommended approach is to remove Enterprise Search data before the upgrade and load it back again afterwards. It significantly reduces the overall operation time.
Limitation: After you remove data, it's not possible to restore ratings and comments added in ISES.
Important
This recommendation applies only while upgrading 11.7.0.1 to version 11.7.1.0.
If you're upgrading from 11.7 Fix Pack 2, it's not necessary to remove and load data, because the data processing speed is significantly higher (approximately 1 million assets per hour).
Before upgrade
1. Log in as the ISES user to the ISES host.
2. Stop services that use Enterprise Search data by running this command. It might take up to a couple of minutes.
kubectl scale --replicas=0 sts shop4info-event-consumer shop4info-scheduler shop4info-rest shop4info-type-registry
3. Verify that all these services were stopped by checking that the output of this command is empty:
kubectl get pod | grep shop4info
4. Enter the gremlin-console command line:
kubectl exec -t -i `kubectl get pods | grep gremlin-console | awk '{print $1}'` -- /opt/gremlin-console/bin/gremlin.sh
The prompt changes to:
gremlin>
5. Delete the existing Enterprise Search graph:
gremlin> g = JanusGraphFactory.open('conf/storage-backend.properties')
==>standardjanusgraph[cassandra:[cassandra]]
gremlin> JanusGraphFactory.drop(g) ==>null
Wait until the operation is finished and
null is displayed as in the example.6. Exit the gremlin-console command line by pressing Ctrl+D. Your operating system prompt appears again.
After upgrade
1. Review recommendations regarding graph synchronization performance: Configure Information Server Enterprise Search for graph synchronization performance and apply them if hardware resources are available.
2. Manually synchronize assets, as described in Synchronizing assets manually.
When ISES is configured for graph synchronization performance, the manual synchronization is approximately 15 times faster than ISES data processing during an in-place upgrade. Information Server is fully functional while the manual graph synchronization is in progress.
Document Location
Worldwide
[{"Business Unit":{"code":"BU059","label":"IBM Software w\/o TPS"},"Product":{"code":"SSZJPZ","label":"IBM InfoSphere Information Server"},"Component":"","Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"11.7","Edition":"","Line of Business":{"code":"LOB10","label":"Data and AI"}}]
Was this topic helpful?
Document Information
Modified date:
22 October 2019
UID
ibm10880705