It comes to the Mandalay Bay Convention Center in Las Vegas for a few days in May. Each year, it gets more complex, and more daunting. It serves thousands of users for a short - but critical - time frame. And the man responsible for it is Glenn Evans, Chief Engineer of the InteropNet. We interviewed him about this massive project, and this is what he had to say:
Q: How long have you been doing InteropNet?
A: I started as a volunteer back in the mid-90s in Australia. Then I moved over to take charge of the InteropNet project in 2004. Then I returned for a second stint in 2010. So I've been doing it for 15 years now.
Q: What has been the most surprised you've been, running the InteropNet?
A: I'm surprised it actually works sometimes. I'm sort of semi-joking there, but when you think about it, I've got 20 vendors, 100 people, and multiple thousand attendees. In general, we don't get that many problems. You could say its good planning, good execution, but sometimes it comes as a bit of a surprise.
Q: Enterprise data center deployments can take months or years. How do you get InteropNet up and running so quickly?
A: It may appear so, but that's not really the case. In reality, from initial concept, to design, to implementation, there's around about a six month process.
Our cycle starts with an RFP that goes out to all the vendors approximately six, maybe seven months ahead of the scheduled show date. And in that RFP there's essentially a concept of what we're trying to achieve for the upcoming show. We take the key points, ('this is what we're looking for, these are some of the things we'd like to achieve') or the key topics of that particular show and then apply InteropNet design or the RFP against that.
Once we then get responses from the RFPs, we then look at the vendor proposals and see how they match the design concept or goals. Can we fit them in, if at all? If we have one or two vendors that can do the same thing, can we get them both in the mix? And what sort of demarcation can we use to get both those vendors in?
When we've made those selections, we then move into the design and the planning phase. We essentially get everybody in a room together and say: 'Guys, here's the concept, here's our deliverable, how do we get from concept to delivery?'
We then take on ideas from all the particular vendors. Certainly in the networking space, while there's a lot of commonality, there's also a lot of proprietary ways to do things, and we try to match them all together.
This type of design and planning phase takes two or three months. And that's basically taking best practices from all the vendors, and fitting it together like a bit of a jigsaw puzzle.
We then move into getting the equipment shipped to the warehouse and start to move from paper design into active electronics, wire, and cable. We test all that, break it, fix it, pack it up, send it to the show, and reinstall it again.
That's the lifecycle we use. And I suppose, to some degree, it's come out of having done this for so long. The InteropNet's been around for 25 years, so we've learned some things over the years to speed up or streamline the process. But we're slightly different in that we're trying to put twenty vendors together, and we've got to take all the different models and see how they fit together with the enterprise, whereas most datacenters will standardize on a particular vendor or a couple of vendors.
Q: How is InteropNet 2011 different from previous years?
A: This is the 25th year of the Interop show, and Interop started out as being a plug-fest, to allow vendors to test interoperability of disparate equipment. Over time, the standards have evolved and a lot of the early networking technologies would fall by the wayside. And the Ethernet became the dominant transport mechanism.
Interop then became more about showing newer technologies or applications that could sit on the underlying transport mechanism. For the last ten years, as part of that, we've tended to have a single-vendor model where one vendor would do all the networking for the whole show, and one vendor would do all the management, etc.
For 2011, we've gone back in time to some degree, where we have multiple vendors doing some of the various aspects or elements of the show. So Cisco, Vyatta, and HP are providing network switching and routing. We've got a number of different management platforms available to us. We have several packet analysis and monitoring vendors.
Q: What has been the most difficult choice you've had to make this year?
A: Funnily enough, in general, my difficult choices usually come about from some of the 'hotter' or 'more vibrant' technologies and networking concepts. This time around, some of the biggest challenges have been centered on the network monitoring, management, and data capture area.
It's a fairly broad sort of area, and a broad scope within the networking area, but it covers people like the tap vendors, packet-analysis, and management platforms, both SNMP and sFlow/NetFlow. My take away from that is that: 'If I've got a difficult choice, there's obviously a lot of innovation happening in that particular area.'
Q: What are you most excited to be showcasing this year?
A: To be honest, it's the overall thing. We're doing the multi-vendor thing. It's a real-world scenario, and it's still of interest to us seeing a few people. Certainly, I'm looking forward to the IPv6 deployment scenarios and the education options we have available to the attendees. To me, that's an exciting thing, because the InteropNet is a demonstration and educational system.
It's a challenge for us - but it's an exciting challenge for us.
Also, [I'm excited about] the use of more intelligent metering and metric systems. That's more from an operational perspective. We're now able to get real-time statistics on NetFlow and sFlow, and that's through the ScienceLogic platform. But we're also doing the traditional SNMP and syslog type monitoring.
'There's also the voice and video solutions. We've got some new video stuff from Cisco, from the Tandberg acquisition that they're presenting. So we'll have some tele-presence systems positioned around the event.
I think the other exciting thing is the vendor mix in the "co-los" [Colocated datacenters] doing the routing. We've got Cisco, Vyatta, and HP. This gives us the ability to show how open source can live with the other guys, in a real-world environment.
Q: What is one of the most nerve-wracking problems you've had?
A: Years ago we had a static problem in Las Vegas Convention Center, and every time we moved a copper cable, it would damage the switching environment. So we ended up having to rewire the switches with anti-static ports the night before the show floor was due to open. That was a challenging moment.
Q: Have there ever been any fatalities associated with InteropNet?
A: We've killed some equipment, but the people have been fine'.
Q: How long have you been doing InteropNet?
A: I started as a volunteer back in the mid-90s in Australia. Then I moved over to take charge of the InteropNet project in 2004. Then I returned for a second stint in 2010. So I've been doing it for 15 years now.
Q: What has been the most surprised you've been, running the InteropNet?
A: I'm surprised it actually works sometimes. I'm sort of semi-joking there, but when you think about it, I've got 20 vendors, 100 people, and multiple thousand attendees. In general, we don't get that many problems. You could say its good planning, good execution, but sometimes it comes as a bit of a surprise.
Q: Enterprise data center deployments can take months or years. How do you get InteropNet up and running so quickly?
A: It may appear so, but that's not really the case. In reality, from initial concept, to design, to implementation, there's around about a six month process.
Our cycle starts with an RFP that goes out to all the vendors approximately six, maybe seven months ahead of the scheduled show date. And in that RFP there's essentially a concept of what we're trying to achieve for the upcoming show. We take the key points, ('this is what we're looking for, these are some of the things we'd like to achieve') or the key topics of that particular show and then apply InteropNet design or the RFP against that.
Once we then get responses from the RFPs, we then look at the vendor proposals and see how they match the design concept or goals. Can we fit them in, if at all? If we have one or two vendors that can do the same thing, can we get them both in the mix? And what sort of demarcation can we use to get both those vendors in?
When we've made those selections, we then move into the design and the planning phase. We essentially get everybody in a room together and say: 'Guys, here's the concept, here's our deliverable, how do we get from concept to delivery?'
We then take on ideas from all the particular vendors. Certainly in the networking space, while there's a lot of commonality, there's also a lot of proprietary ways to do things, and we try to match them all together.
This type of design and planning phase takes two or three months. And that's basically taking best practices from all the vendors, and fitting it together like a bit of a jigsaw puzzle.
We then move into getting the equipment shipped to the warehouse and start to move from paper design into active electronics, wire, and cable. We test all that, break it, fix it, pack it up, send it to the show, and reinstall it again.
That's the lifecycle we use. And I suppose, to some degree, it's come out of having done this for so long. The InteropNet's been around for 25 years, so we've learned some things over the years to speed up or streamline the process. But we're slightly different in that we're trying to put twenty vendors together, and we've got to take all the different models and see how they fit together with the enterprise, whereas most datacenters will standardize on a particular vendor or a couple of vendors.
Q: How is InteropNet 2011 different from previous years?
A: This is the 25th year of the Interop show, and Interop started out as being a plug-fest, to allow vendors to test interoperability of disparate equipment. Over time, the standards have evolved and a lot of the early networking technologies would fall by the wayside. And the Ethernet became the dominant transport mechanism.
Interop then became more about showing newer technologies or applications that could sit on the underlying transport mechanism. For the last ten years, as part of that, we've tended to have a single-vendor model where one vendor would do all the networking for the whole show, and one vendor would do all the management, etc.
For 2011, we've gone back in time to some degree, where we have multiple vendors doing some of the various aspects or elements of the show. So Cisco, Vyatta, and HP are providing network switching and routing. We've got a number of different management platforms available to us. We have several packet analysis and monitoring vendors.
Q: What has been the most difficult choice you've had to make this year?
A: Funnily enough, in general, my difficult choices usually come about from some of the 'hotter' or 'more vibrant' technologies and networking concepts. This time around, some of the biggest challenges have been centered on the network monitoring, management, and data capture area.
It's a fairly broad sort of area, and a broad scope within the networking area, but it covers people like the tap vendors, packet-analysis, and management platforms, both SNMP and sFlow/NetFlow. My take away from that is that: 'If I've got a difficult choice, there's obviously a lot of innovation happening in that particular area.'
Q: What are you most excited to be showcasing this year?
A: To be honest, it's the overall thing. We're doing the multi-vendor thing. It's a real-world scenario, and it's still of interest to us seeing a few people. Certainly, I'm looking forward to the IPv6 deployment scenarios and the education options we have available to the attendees. To me, that's an exciting thing, because the InteropNet is a demonstration and educational system.
It's a challenge for us - but it's an exciting challenge for us.
Also, [I'm excited about] the use of more intelligent metering and metric systems. That's more from an operational perspective. We're now able to get real-time statistics on NetFlow and sFlow, and that's through the ScienceLogic platform. But we're also doing the traditional SNMP and syslog type monitoring.
'There's also the voice and video solutions. We've got some new video stuff from Cisco, from the Tandberg acquisition that they're presenting. So we'll have some tele-presence systems positioned around the event.
I think the other exciting thing is the vendor mix in the "co-los" [Colocated datacenters] doing the routing. We've got Cisco, Vyatta, and HP. This gives us the ability to show how open source can live with the other guys, in a real-world environment.
Q: What is one of the most nerve-wracking problems you've had?
A: Years ago we had a static problem in Las Vegas Convention Center, and every time we moved a copper cable, it would damage the switching environment. So we ended up having to rewire the switches with anti-static ports the night before the show floor was due to open. That was a challenging moment.
Q: Have there ever been any fatalities associated with InteropNet?
A: We've killed some equipment, but the people have been fine'.
Source...