Hi Dojo,
I’ve got a requirement for faster alarm search in a (heavily distributed) DMS.
Together with the short term approach for alarm reduction, the longer term strategy would consider ElasticSearch too, for an optimized text search in the DBs (where Cassandra is more focused on the amount of data, I understand Elastic can make text search faster in the NoSQL DB – but please advise if got this wrong).
Now, within this scenario, I understand that ElasticSearch is indeed recommended in presence of a Cassandra cluster – is it the only case where Elastic can be used? That is.. is a Cassanra cluster always a prerequisite of an ElasticSearch cluster for DataMiner?
Or could a DMA be dimensioned to have Cassandra & Elastic locally?
Not that I would do that, but just to better understand the course of action & the possible steps to move from a traditional “local DB” architecture towards the flexibility achievable with Cassandra & ElasticSearch clusters.
Thanks
Alberto, you will see all the supported architectures recommended by Skyline at Supported system data storage architectures | DataMiner Docs .
Many features are distributed across Cassandra and Elasticsearch, however as you mentioned Cassandra is more to manage the huge volumes of trend data and where as Elasticsearch excels in managing the text data like alarms. Many of the new feature including SRM relies on the Elasticsearch storage, so you need to consider how you will be using the DataMiner in the field.
As you will see in our documentation, we only recommend the local DB option for DEV environments but not for the production. Architectures without Elasticsearch and also using the local Cassandra instead of Cassandra cluster are supported but these are not recommended by Skyline for the new deployments.
Thanks.
Thanks for the link, Srikanth – much appreciated!
Without official reference, I would have been like the Cassandra of the Greek mythology 😉 – wishing you a great evening
PS: Would there be any measurement available in terms of how faster an alarm search becomes in a DMS after that an ElasticSearch solution has been deployed?