Ask the Experts: Designing Private Clouds for Applications

Welcome to Ask the Experts, brought to you by In this video, Intelisys’ SVP Cloud Transformation Andrew Pryfogle talks about designing private clouds for customers’ applications with UnitedLayer’s CEO Abhijit Phanse. Find out more about private networking and cloud services from Abhijit and the UnitedLayer team here:

Andrew: All right, guys. Time to jump into another Ask the Experts session. I love this guy I’m about to introduce. He’s one of the smartest guys I know in this space, one of our go-to smart guys for the University. I want to welcome him to the studio once again. Abhijit Phanse, CEO of UnitedLayer out of San Francisco. Abhijit, great to see you again, man.
Abhijit: Great to see you, Andrew.
Andrew: All right. Very cool. Thanks for jumping in. You know, we’ve been talking about … in this latest certification track–this Advanced Data Networking certification track–we’ve been talking a lot about applications. You know, when we design a WAN successfully, the questions around the applications that a customer is using are so critical to a successful outcome. It also is true for a cloud outcome.
I wanted to get your perspective. When you’re designing a solution for a client, what types of applications are performing really, really well in a cloud deployment? What are the behavioral characteristics of some applications that you really need to understand in order to have a really positive outcome for a customer? What do you think about that?
Abhijit: Andrew, that’s a great question because I think almost every application is ideal for some kind of cloud. Not the same cloud, but some kind of cloud. Especially a cloud, Andrew, which has been optimized for that application. It’s a fantastic question. Let me give you some case studies. We are a hybrid cloud company and we architect private clouds for a given application. We understand its requirements, specifically its latency requirements, which is a key consideration while designing the LAN, the WAN, and the network.
I’ll give you a few examples. We’ve got a client, which is a large social media client, and they have to put out a lot of video content, photographs. They want to be very close to the end user, so the end user has a great experience. Out here, we park the cloud very close to where the user base is to make it a really low-latency experience for them.
We have another large media company which puts a lot of movie reviews online. Again, same principle: very low latency, close to the end customer.
Then we’ve got a customer which has a cloud which we’ve deployed. It’s a Hadoop private cloud for determining on how to bid on auctions for advertisements. Let’s say, Andrew, you’re actually shopping for a new Audi. The cookies are figuring that out, so when go to CNN, they follow you. Right? How do they do that?
There’s an engine behind which is figuring out a marketplace that Andrew Pryfogle wants to buy an Audi. An Audi, let’s say, in San Francisco wants to sell that to you. They bid on that ad to be posted for you. That has to be done in real-time because if they’re late in making that bid, they lose that bid to somebody else.
It’s a very latency-sensitive application. Again, that has to be architected understanding the compute horsepower, the latency in the compute, but also the network and the user, and the experience from a latency perspective for the end user.
The last example I was going to use was one of our customers, Wikipedia, is one of the largest content distribution companies in the world today; largest traffic in the world today; one of the top three traffic companies in the world today. Again, we designed a very high quality, high performance peering fabric to drive a very low-latency experience for the end user base.
Again, lots of examples where you would actually design your private cloud to meet the end users’ application needs.
Andrew: Yeah. Very cool. It sounds like applications or understanding the application use of a customer is critically important to a successful outcome.
Abhijit: Absolutely, Andrew.
Andrew: Not every cloud, not every network is going to handle every application successfully. Is that a fair statement?
Abhijit: Absolutely. I’ll give you another example, which is a different example, which makes that point. Disaster recovery, big application. Very different kind of needs. You want it to be far away, but sometimes you don’t want it to be too far away. We’ve got a customer which is a VoIP provider and they will want to be 100 miles away so if disaster hits in their primary cloud, they can actually go to another site. If they’re too far away, the latency to get to that cloud infrastructure is going to hurt the VoIP quality.
It’s fine-tuning the needs of the customer and the needs of that application while designing your WAN.
Andrew: Very cool. That’s great insight, Abhijit. Thank you very much, man. I appreciate that.
Guys, that’s Abhijit Phanse. He’s the CEO of UnitedLayer out of San Francisco, one of our go-to cloud providers in the portfolio. Again, one of the smartest guys that I know. Abhijit, thanks for jumping in, man.
Abhijit: Thank you, Andrew.
Andrew: Hey guys, do make sure you check out the learning center for UnitedLayer right here at the University. It’s chock-full of great information, great case studies, and videos, and white papers–sales tools that you can use to close more deals in the cloud, especially with UnitedLayer. Good selling.