As I was reviewing the slides from Domino Strategy V11 deck from the Factory Tour 3 HCL Presentation Catalog posted here, I was thinking about some use-case scenarios with DAOS in AWS S3.
Wait… DAOS in S3? Somehow I had missed a reference to DAOS being able to integrate with cloud in past announcements, but this has a lot of potential….
Since DAOS is really WORM storage (write once read many), it may make sense to keep deduplicated attachments up in the cloud. But in some cases it may not, so YMMV. Before we look at a couple of options and scenarios, here is the relevant slide.
If you are not familiar with S3, it is far more than just cloud file storage. You can also have an on-prem AWS Storage Gateway which is basically a VM that caches the most used objects locally for faster access. Calls to other objects in S3 would go out to the cloud. AWS also has many regions. Depending on your location, you may have faster transfer speeds to a region that you wouldn’t think would be better. It’s not just a glance at a region map and picking which one is closest, though that is obviously a factor. You can test out transfers to different regions here. One other thing that is available is Cross-Region Replication. This is what you think it is. 😉 Also, there are various S3 storage tiers with different levels of availability and redundancy.
So let’s take a quick look at some DAOS scenarios and you’ll see where cloud DOAS object storage can possibly benefit you.
- You have a mission-critical Notes app which may use large attachments that are not best downloaded over a WAN link. However, older data is not referenced as much. Maybe you could have the Storage Gateway so that retrieval of newer data is faster but then everything is still out in the cloud.
- You open a new office on the other side of the country or in another country and need to quickly provision them a Domino server to let them use the mission-critical app. Since the attachments can be shared across servers, you could enable Cross-Region Replication and then replicate the NSF’s over to the new site more quickly.
- Consider having a hot clustermate out running in another AWS Region in case you have an outage. This will also let you spin that up faster.
- You also now have redundant offsite storage of your attachments and can likely recover fairly quickly from an outage. Especially if you may have chosen to also backup (or store copies of NSF backups) out on S3 in another, cheaper storage tier.
- Oh, and if you’re worried about getting possibly terabytes of DAOS objects up to AWS, just throw them a Snowball. 🙂
These are just a few scenarios off the top of my head. What about you? Do you have ways you’ve thought about leveraging this feature? It’s not too soon to begin planning.