#community-help

Deploying and Scaling Typesense on Digital Ocean

TLDR Bill guides Kishore Nallan on deploying Typesense on Digital Ocean App platform and discusses scaling options. Kishore Nallan recommends vertical scaling and provides hardware recommendations for Bill's specific workload.

Powered by Struct AI

1

12
21mo
Solved
Join the chat
Apr 03, 2022 (21 months ago)
Bill
Photo of md5-be53735a2b0297bb542711c1d2ecea45
Bill
11:32 AM
Hello, is it possible to deploy typesense in Digital Ocean App platform? I'm not sure because instances can be destroyed
Kishore Nallan
Photo of md5-4e872368b2b2668460205b409e95c2ea
Kishore Nallan
11:35 AM
You can deploy it, but clustering requires private network to be reliable. I'm not sure if Digital Ocean provides an internal network for communication between instances. Other cloud providers like AWS & GCP do.
Bill
Photo of md5-be53735a2b0297bb542711c1d2ecea45
Bill
11:37 AM
Kishore Nallan As I have informed you, I have created a multi node cluster (Load balancer with 3 droplets - Ubuntu) but I want to find a way in order to scale my setup fast. For example, in app platform I can set easily my instances to eg. 6. Do you suggest any other way for scalability?
11:39
Bill
11:39 AM
With my setup now, if I want to add 2 more nodes, I have to create 2 new droplets, install typesense in ubuntu, set nodes etc
Kishore Nallan
Photo of md5-4e872368b2b2668460205b409e95c2ea
Kishore Nallan
11:47 AM
Any reason for not scaling vertically, rather than adding more machines? Since Typesense does not shard the underlying data, vertical scaling should work well.
Bill
Photo of md5-be53735a2b0297bb542711c1d2ecea45
Bill
11:52 AM
So, it's better for example to have 5 nodes and scale them vertically? What min size (RAM-CPU) of instance do you recommend for production?
Kishore Nallan
Photo of md5-4e872368b2b2668460205b409e95c2ea
Kishore Nallan
11:57 AM
Yes, as you add more nodes, write latency will increase. Exact RAM/CPU depends on your dataset and traffic volume.
Bill
Photo of md5-be53735a2b0297bb542711c1d2ecea45
Bill
11:58 AM
My dataset will be static, on average 3-5K docs max. Do i need a cpu optimized instance or more ram?
Kishore Nallan
Photo of md5-4e872368b2b2668460205b409e95c2ea
Kishore Nallan
12:03 PM
You don't need much RAM for that. How many queries / second are you expecting during peak traffic?
Bill
Photo of md5-be53735a2b0297bb542711c1d2ecea45
Bill
12:05 PM
50-100
Kishore Nallan
Photo of md5-4e872368b2b2668460205b409e95c2ea
Kishore Nallan
12:07 PM
5 nodes with 2 CPUs each is sufficient to handle that provided the requests are evenly distributed across all nodes. That's a good starting point. You can check the CPU usage and decide if you need to switch to either 3 node with 4 CPU or 5 node with 4 CPU after that.
Bill
Photo of md5-be53735a2b0297bb542711c1d2ecea45
Bill
12:09 PM
Okay thank you Kishore 👍

1

Typesense

Lightning-fast, open source search engine for everyone | Knowledge Base powered by Struct.AI

Indexed 3005 threads (79% resolved)

Join Our Community