CTO Daily Dose - How cloud hotels help with multi-cloud
Transcript
hi Keith bouncer from the CTO visor calm I know you guys looking at it another funny camera angle we're working these things out hey this is I got a give a tip of the cap to Missy young over a switch this is inspired by her recent news that she's now CIA over a switch congratulations Missy nice to see women in stem moving forward and I'm sure it's because of just a lot of hard work however it reminded me of a conversation I had over on the
podcast with Mark teal some months ago when he was over s switch and that's the enabling force that a hotel that a cloud hotel ur is so that term is a little really weird and I think will learn what it is in a bit so one of the interesting kind of movements in enterprise IT is to say you know what we don't want to just go with AWS or Azure or Google compute we want a moti cloud strategy I talked to a really large global firm
the other day and we were talking through the challenges of multi cloud and one of the biggest challenges is latency so we've talked about this over on the cube John for your likes to refer to this as the latency problem or the speed of light problem let's draw that out really simple so you have two cloud providers let's say you have Google GCP and Azure you want to build connectivity in an app that let's say you want to take advantage of GCPs ai they have really
great AI and Azure art obviously has strong Linux and windows and past support you want to build an application that spans these two services one of the obvious things you can do is create a VPN between the two however especially with heavy data analysis the latency between the two physical cloud providers may prove to be too great to actually have a functioning application so what's some of the solutions to that there's a bunch but let's talk specifically about where whole cloud hotel comes into play this
is also referred to as a CNF cloud I'm sorry carrier neutral facilities in which you have someone like a equal mix a switch that are basically MeetMe points so when you go to let's say AWS and say you want a Direct Connect AWS isn't gonna give you a circuit straight to your data center they're gonna meet you at one of these cloud hotels ie equal Nicks or switch and when they're all put network their Colo network equipment in one of these facilities you'd call it network
equipment in those facilities and then you can basically have anywhere from a 1 gig to a 2 gig - I'm sorry - a 10 gig connection between your facilities and the Colo and the facilities at Google compute so if your data center is down here you can have a one gig MPLS circuit direct can or point to point and you'll virtually have a one gig connection to your cloud providers instead of having multiple connections to each cloud provider you have a single connection and you have
a one gig connection to the cloud the great thing about this is that we can add yet another cloud provider now we have one gig throughput to any cloud provider this makes building applications on-premises and talking to cloud providers great but it doesn't solve the legacy problem no matter how much bandwidth I have I have I can upgrade this link to ten gig and it will still take 70 milliseconds to get from here to here to here or to their eyes and that's an incredible amount
of latency when I have sup one millisecond latency from one my datacenter will fit in my data center so building cross cloud applications becomes tough in this environment however what we can do we can take the solution like clear sky clear sky does some really interesting things and this is the simple way to do it you can outsource the clear sky you can look at a solution like NetApp rubric again a ton of people do this from either a secondary or primary data perspective in which
there's a device primary storage living in the equinox or switch facility and our CNF or there's a service living here with a caching layer at our on premises data center so virtually I have that super low data center speed quality from my on-premises solution into the cloud provider or the CNF the hotel and then my cloud providers or my cloud services can vent access that same storage layer at super low latency again we can put these accelerating acceleration devices inside the cloud provider I think there
are several players they'll let you do virtual appliances and the cloud so now I've created this super low latency network data network across multiple cloud providers leveraging both physical high-speed connections into a data center hotel or cloud hotels CNF and the adding a layer of caching devices on top of that and physical storage to provide a really robust low latency network now I haven't seen that's done really well this field I haven't talked to a customer yet this implemented this so I still have a few
questions about how this functionally works you know is this great for transactional data I can see some problems that you're trying to run transactional workloads across multiple clouds that low that that caching layer may not be up to snuff I don't know I haven't seen this done at scale but this is conceptionally how a cloud hotel can help in your multi cloud strategy that's it for this CTO adviser will are the CTO daily dose we'll talk to you next daily dose again if you'd like to
sponsor to the daily dose your message could have been at the beginning of this episode you can find out more about sponsorship opportunities sales at CTO advisor comm we can have this discussion if you'd like to talk more about this on LinkedIn that's my favorite medium for going back and forth this video will be posted on LinkedIn look me up on LinkedIn or you can just search for the hashtag hashtag daily dose talk to you guys next daily dose
