Good Morning - We are looking at setting up Typese...
# community-help
r
Good Morning - We are looking at setting up Typesense in GCP with HA. The source of the data would be a DWH and an OLTP - and we would write APIs that would leverage Typesense APIs to continuously ingest data. If we setup HA by ourselves, would we need to ingest data into two zones if we want to implement HA by ourselves as we are not able to find any replication across zones? We also see documentation about MongoDB integration with Typesense and Firebase with Typesense. Assuming that these would be available for any document based ingestion with the APIs provided, can we leverage the managed services in GCP?
j
i believe typesense does the replication itself as long as you have 3+ typesense nodes...they just all need to be able to reach each other over the availability zones for mongo, as long as the managed mongo supports change streams that are accessible to your typesense nodes, it should be fine: https://typesense.org/docs/guide/mongodb-full-text-search.html#step-2-start-a-mongodb-replica-set
2
k
👋 @Rengarajan Deivanayagam
If we setup HA by ourselves, would we need to ingest data into two zones if we want to implement HA by ourselves as we are not able to find any replication across zones?
You will just set up a cluster whose nodes that are across regions. You don't have to do anything special from Typesense perspective. There might be additional latency but Typesense will handle it.
Assuming that these would be available for any document based ingestion with the APIs provided, can we leverage the managed services in GCP?
I don't quite understand this question. Can you please elaborate? Are you using Mongo as the primary data store?
r
@Kishore Nallan thanks for the quick response. My second question is about the managed services and whether we could ingest data from sources other than mongo or firebase such as SAP HANA or Amazon Redshift (through a middleware)
@Joshua Hoover thanks will check that out. And no we won't be using mongo, but have found ways to generate files that can be continuously ingested
👍 1
k
@Rengarajan Deivanayagam Given the intricacies/options involved in data ingestion, maybe easier to discuss this over a call? Depending on your timezone either @Jason Bosco or I would be happy to do a quick call.
r
Sure thanks @Kishore Nallan . I am available in IST. Would sometime tomorrow be fine - or would Monday be ok? I can send an invite or can jump on a call
k
Tomorrow is fine @Rengarajan Deivanayagam Here's my calendar: https://calendly.com/kishorenc/30min Feel free to pick a time that suites.
r
Thanks @Kishore Nallan will setup one soon.
👍 1
k
@Rengarajan Deivanayagam I've a scheduling conflict that just came up in the evening tomorrow. I will be away from 5 pm to 8.30 pm. You free any time between noon till 5 pm? I've DMed you