CTO Daily Dose - Disaster Recovery in the Cloud
Transcript
getting started with today's topic which is dr in the cloud I think I think this topic came from my good friend Sunday or on I'm not sure it was LinkedIn or Twitter but if you want to suggest the CTO daily dose hashtag CTO daily dose and we'll see if we can fit the top again but today's topic is dr in the cloud and whether or not one we need dr the cloud and if we do need the artist out how do we do it so let's
talk about what we do to in today's data center so whether we do this at the application layer or at the storage layer the basic concept is that we have two types their primary data center secondary data center we can get fancy and say that we have an active passive active active that's not really relevant what but the basic concept is that we have some type of replication whether it's storage replication or application replication where we get the data from a primary data center let's say
I'm in Chicago and I want geographic protection so I might have a data center on the west coast so in California and San Francisco and replicate data from Chicago to San Francisco if my primary data center fails then I can fell over all of my data and applications are stored in San Francisco the big question becomes well do we want to do this in the cloud and if so how does it look we get really complex and the physical design of this set up in the
cloud all of that is abstracted away we don't have the concept of a primary data center most cloud providers have this concept of an availability zone or a region so within a region we may have a conceptionally we might have let's take a ws for example we'll have the Virginia region and then within the Virginia reason we might have several availability zones within the Virginia region and that availability zone we might have a primary set of data we can replicate that data with fencing the same
availability zone to other availability zones so that we have tertiary or secondary copies of our system data again this can be done at the simply by saying we're going to replicate the object storage from one availability zone to another availability zone or we're going to do application level replication same technical approach that we would take in the data center the question is why why would we even want to do that this is Amazon they stay up all the time right not necessarily the Amazon has different
availability zones within a region because something can physically happen to one of those data centers within that availability song zone let's say that power is lost in the primary availability zone and you didn't replicate your or dispose your rope-work little across availability zones and now you're just down the the region maybe up as a whole but the one availability zone within that region is up is down I don't consider that disaster recovery that's more of a H a solution within your data center you might have
a super important Oracle database so in your data center you might cluster that database across multiple holes so that if a physical hole fit one physical host fails your database phase available that's the same concept basically than one availability zone or in one data center if availability zone fills within that region then your application stays up pretty same pretty similar concept what we want to think about is what's the secondary location at which we have just in case this entire region fails this is where we
need to think about what it is our business objective is is our business objective for our business continue ality because disaster recovery is just a subset of business continued allottee what is the business risk or the environmental risk that we're trying to protect against are we trying to protect against Amazon's physical region going down which is a technical concern and that could be amongst a list of things that we want to protect against or are we trying to protect against Amazon going down meaning Amazon itself
not being available for whatever reason the company goes out of business there is a business dispute and you and Amazon shuts down your systems or you need to evacuate your data out of AWS as soon as possible so based on that you have two options you can go to another region within AWS so right now we have our primary data center is the VA Virginia data center and then we might decide to do Oregon as our secondary region within the AWS Network we can use the
same basic technologies we can do objects ec2 instances and Oregon that mimic our easy to images in in Virginia we can do object store s3 replication from one available T zone or one region to another region or we can go completely off the beaten trail and say that we're making a business decision to protect against Amazon's environment and in general and choose the goal to Oregon in in something like a zoo that is much more complicated from just a technical and cost perspective than doing this
but offers additional protection so this the picture looks a lot similar to our original picture when we're doing physical data center replication we can now do storage type replication which there's appliances that allow us to do this I think one example is rubric will let you back up into you know f3 format and then then Chris wall will correct me if I'm wrong and then back up into Azure there are several products that allow you to do storage based replication from one service provider to a
different service provider to give you that service provider level protection and then you can also do it in application you didn't just build your application and say that you know what replicate data from this target to another target so you have a set of instances let's say you have Windows instances running in Amazon and Virginia East and you have equivalent images of course different underlying infrastructure running in the zoo and you just provide IP connectivity between the two and they you replicate data in between so
dr doesn't go away just because you have cloud we can't outsource the responsibility of business continued allottee to our cloud provider let's feel a core accountable accountable countability perspective that's and responsibility upon your existing IT team we have to figure out how we're going to reach the business objective of either protecting against the business relationship or systems within a cloud provider failing and ensuring that we have the appropriate risk mitigation whether it's going to another cloud provider or going within the same cloud provider to offer
some level of high availability or durability to our data that's it for this CTO daily dose follow me on the web at CTU advisor the website is WWE TV khmer the blog these videos and as well as the podcast talk to you guys in tomorrow's Daily No
