Now that we have seen how flexible it is, to scale out storage not only in the underlying AOS cluster but Objects can scale across multiple AOS clusters, breaking the physical boundary .
But if your workload is CPU/Memory intensive, e.g- if your application is writing/reading small files in high number , then your IOPs (w.r.t number of Gets/Puts/(S3-requests)) are bound by amount of compute available to your Objects cluster.
In this case, you may want to expand your Objects cluster to get better IOPs.
With Objects2.1, Single Click scale-out is supported right from Objects UI.
Navigate to PC -> Objects UI -> Select Objects Instance -> Click on “Add Node”
This will take about 5-6m (Little +/-), and your Objects cluster is scaled by one more Objects node. To add multiple nodes, you can repeat the above action.
Above screenshot also gives you an option to update storage-capacity which is optional. Useful from a licensing perspective , once given storage capacity is reached, you will get notification in “Alerts Tab” of your Objects cluster. I am increasing as well.
It’s that Simple.
Few things to Note here :
- You can scale-out your Objects cluster as big as the underlying AOS cluster. I.e if you have an 8N AOS cluster, then you can scale out your Objects cluster upto 8Nodes.
- It's recommended to have N+1 AOS Nodes for HA purpose. i.e To deploy 4N Objects cluster, you should have 5N AOS cluster. So in above example, you should scale your Objects cluster to 7N on 8N AOS cluster.
- You can not scale-out One or two node Objects cluster . But the Multi-cluster feature is available for all types of Objects cluster.
- With addition of each node, Objects will take up additional 32G memory and 10 vCPUs .
- You also need additional IPs in your network which is used for deploying Objects cluster.
Give it a try, and be amazed .