As I was explaining Gladinet CloudAFS to a group of IT people as a file server with cloud storage attached, one question I got was:
How much bandwidth you need to run file server with cloud storage?
It is very obvious where the concern is coming from. If I had an on-premise file server, with cloud storage attached as tier 2 storage, will the slowness of the Internet (as compared to LAN) slow down the file server activities?
Rest assured, Gladinet CloudAFS takes care of this. Specifically, it takes care of two issues.
A. Slow Internet to the Cloud Storage – that is why it is tier2 storage.
B. There is a glitch, Internet is down.
The secret to both maintain local LAN performance and leverage Cloud Storage is to have a tier 1 storage locally on the hard drive (or NAS or SAN). When people read and write to the file server, it is writing to and reading from the tier 1 storage, thus gives you an impression of local storage because it is indeed local storage.
Gladinet CloudAFS will then schedule transfer to and from the tier 2 storage asynchronously. You can schedule the transfer to be 12 hours later than the tier 1 storage hit so it can happen at night when nobody is working. If the Internet is down at a moment, the transfer will resume and retry later on.
Use Case 1 – File Server Backup
When your Internet is slow or the Internet is not quite reliable, you need to put cloud storage to a more secondary role, such as backup.
In this use case, you can configure Gladinet CloudAFS tier 1 storage to have the same size as tier2 cloud storage. This way, the read/write is always from local and you are guaranteed to have 100% up time and 100% local performance. The cloud storage is something nice to have just in case that your local server crash or out of capacity, you can get it back immediately by provision another CloudAFS server (5 minute install when OS is ready).
The benefit of this use case is that you can always quickly re-provision your file server in case something happens to it.
Use Case 2 – Cloud Storage Gateway
When your Internet is fast and the Internet is reliable, you can put cloud storage to a more prominent role, or on the other hand, put CloudAFS to a more secondary role such as a Cloud Storage Gateway with cache.
For example, if you file server is with Amazon EC2 and your storage is Amazon S3, you know you have fast and reliable connections in between, you have a fit for the Cloud Storage Gateway use case.
Another example, if you are hosting with Peer1 server and also using Peer1’s CloudOne storage, you have a fast and reliable connection in between too. This case fit you as well.
In this case, the local cache on the CloudAFS can be smaller than the actual cloud storage capacity.
When the cache is smaller, you will incur cache miss if a read to the CloudAFS server doesn’t already have the file locally. This case the file needs to be fetched from the cloud storage over the Internet. Since you have fast and reliable connection, this first time and one time cost is OK.
The benefit of this use case is that you can ship a lot of data over to cloud storage. For example, you can turn local backup software to cloud backup software by pointing the backup destination to CloudAFS.
The two biggest parameter for CloudAFS are
A. How big the local cache is. The slower and less reliable the Internet is, the bigger the local cache you will need to assign.
B. Scheduled flush time. CloudAFS use a write-through cache with flush schedule. The slower the Internet you have, the more delay you will need to push the synchronization to night time.
At this time, CloudAFS supports a wide variety of cloud storage providers, including Amazon S3, Windows Azure, Rackspace Cloud Files, EMC Atmos, Mezeo, OpenStack, Nirvanix, AT&T Synaptic Storage, Peer 1 CloudOne and more.
Speed is an important factor when you apply cloud storage to file server. Visit other earlier posts about how to pick a cloud storage providers.