David Trossell [00:00:00]:
The only way in in reality can recover from any cyber attack is having that golden backup, that air gap backup. If you’re transporting data and it’s valuable, then encrypt it. We don’t want to encrypt the data for you. If you encrypt the data back in your systems, you keep the keys back in your system. So they’re away from us. They’re away from the way, and they’re away from the rest of it.

Karissa Breen [00:00:38]:
Joining me today is David Trossell, CEO and CTO from Bridgeworks. And today, we’re discussing how to move data faster over WAN. So, David, thanks for joining and welcome.

David Trossell [00:00:53]:
No. My pleasure. Absolutely lovely to talk to you.

Karissa Breen [00:00:56]:
Okay. So let’s start right there. WAN and acceleration, walk me through it. What whatever comes to mind when I ask you that question because I know it’s obviously a very big area widely spoken about. So really keen to start wherever your mind takes you.

David Trossell [00:01:11]:
I think to understand where we’ve landed up with WAN acceleration, we have to go back to the comm era where we started to have broadband connectivity, which was a huge leap up from dialing in on my, modems every so often. But that also created a need to move data around, but the the networks really weren’t there still. You know, the we’re still down into megabits rather than megabytes of data. One of the things you’ve got to understand there is is the the restrictions on the technology. And it wasn’t really until, you know, the early naughties that we started to get some dedicated ethernet connections between customers and telcos and the rest of it. And but they were still only 10 megabits per second. And we started to see also satellite offices appear as well. Again, you know, trying to communicate that them to the head office is a bit more difficult over a modem.

David Trossell [00:02:11]:
But one of the things that popped up now that lower bandwidth and remote offices was a technology called WAN optimization. Now this is quite clever. I mean, it’s a very good idea. So the idea is you have 2 units at either end of the WAN connection. And rather send in all the data across the WAN, what it did was to compress it, what they used to call deduplication. So and then you would then store that data into a table, which you could reference and pull back later, which is very good. It was, absolutely an excellent idea, which means you could run theoretically higher speeds over very low network connections. It did have a limitation.

David Trossell [00:02:53]:
You know, you had to do all this compression. You had to store all the data either side of it, which was good, but it when you want to move up into the megabits and gigabits performance, it just wouldn’t handle it. It just it just doesn’t go that fast. And that was where we sort of stuck up for the time being until we had a a request from one of our distributors that had a site in the north of Sweden and a site in the south of Sweden inside a mountain. So you can imagine what sort of people they were. And they wanted to connect their backup system across that network. And they had an Ethernet link, but they still couldn’t get the storage area network protocol, which in that case was fiber channel over that link and down into the mountain in south of Sweden. So it came to us and said, yo, can you do anything for us? So we took one of our, what we call, protocol bridges and then put in between what we call WAN acceleration.

David Trossell [00:03:50]:
So this is a process driven by AI, which paralyzes the flow of data across the network. You can’t get rid of latency. It’s a fact of life. You know, the only way you can reduce latency, which is one of the big impacts on performance of TCP IP is you you have to think move things closer together, and that’s and that’s governed by the speed of light. We use the AI in conjunction with a program with Portsmouth University to develop this situation where we can actually fill the pipe up with data. You can’t get rid of the latency. It’s just there. But we and it also managed the packet loss and congestion.

David Trossell [00:04:27]:
And that’s where we renamed it from WAN optimization to WAN acceleration.

Karissa Breen [00:04:31]:
So talking about latency, now, I mean, there’s couple of things that you said is quite interesting. I wanna get into them because often, you would know better than me, David, that people complain a lot about latency, and you were just basically, what you said is we won’t be able to fully remove that. So when you’re speaking to customers, what and the latency sort of conversation chatter comes up, what what are sort of main people’s concerns on that front? Anything you can share?

David Trossell [00:04:58]:
Yeah. If we’re using TCPIP to transfer data, which has been around since DARPA, days in the seventies, it’s a great technology because it makes sure every packet gets there, whether it loses it on the way or not. The idea that everything packet will get there in the end. But this problem with latency is that the speed of light is just not fast enough. That might sound silly, but the speed of light is, what, 300,000 kilometers per second in vacuum. But if you as soon as you squirt that down a piece of optical fiber, you lose a surge. So you’re down to 200 kilometers 200,000 kilometers per second. And if you think that between London and New York is about a 100 milliseconds, that’s quite a long time to go out there with your packet and then to get the acknowledge that that packet has made it to the other end and that that acknowledge to come back into London.

David Trossell [00:05:48]:
So really, you can only do that round trip 10 times a second. So that’s the effect of latency. So, unless unless someone covers up with the ability to move fast move data faster than the speed of light, we’re stuck with the speed of light. So you have to take an another approach, that is that rather than trying to get this get down is to get the the amount of data you transmit up each time. So what we do with the artificial intelligence is create parallel streams across this network connection. So we fill the pipe up with data. So the first byte is gonna take its 100 milliseconds to get there, but all the other bytes are coming along at high speed behind them. So the idea is you move the data as fast as possible across the network by you you’d maximize the utilization of the network.

David Trossell [00:06:37]:
So that’s where we’ve gone from 1 gigabit per second to 2 gigabit per second to 10 gigabits per second, 40 gigabits per second, and we’re now trying out a 100 gigabits per second transfers.

Karissa Breen [00:06:50]:
I wanna go back a moment. So you mentioned the amount of data you transmit. Would you say with your experience, companies don’t understand that phrase that you just mentioned or how to do that effectively to, you know, to reduce the latency, for example?

David Trossell [00:07:09]:
You can’t reduce latency. The only way you can reduce latency is to move the two ends together. That’s the thing that a lot of people don’t understand. That’s why if you go to New York or London and you see the high speed trading, latency is, is it down into microseconds? So they actually put their computers right next to each other. If you see what I mean, that’s the only way you can get latency down. How you can actually get the efficiency of that network link up, say, between New York and London is to is to fill the pipe up with data. So you’re gonna move more data quickly across the network, and you use that parallelization technique to do that. And that’s all run by AI.

David Trossell [00:07:47]:
The AI is quite clever because it will look at the network and say, can I get more on that network, or can I get less on that work? It’s am I getting packet loss, which could be caused by switches or other people getting on the network at the same time. Do I actually change the size of the packets? Do I make the packet smaller? So if I do lose a packet set, I only have a little bit more to reconstitute constitute and push back. And it would do things like so it changes the parameters it has for the network, how the data comes in, and how the data flows out by the two ends. Hopefully, that makes sense.

Karissa Breen [00:08:23]:
Yeah. So okay. So when you’re saying that, would you say that’s probably a big misconception then? That you said you you can’t reduce it. You can’t reduce the latency. So would you say people and when I mean people, I mean companies out there that you’re speaking to, would that be their assumption a lot of the time that you can reduce it, but you’re saying you can’t? Would that be sort of the main issue?

David Trossell [00:08:41]:
That’s right. You can you can mitigate the effects of latency, but you can’t reduce latency unless you move it closer together, unless you run things in a very, very straight line, which you can’t do if you’re going across continent. However, what most people do these days, if they have a, poor, transfer across their networks, they tend to talk to the Telco first. And the first thing the Telco saying is, well, you need a bigger pipe. It’s going to cost you this much and that will make it go faster. But the fatal latency means you don’t get any increase in performance because it still takes that amount of time to get from one end to the other and back again to get back that import all important acknowledgment you have to have before you can send the next set of packets.

Karissa Breen [00:09:26]:
So you said before the only way to really get around it would be to, you know, move it closer, for example. But now with the way the world’s moving with this whole people working from anywhere, travelling, working remotely, like, that as sort of a a solution is probably not the reality, would you say? Because even now, like, people aren’t even going to the office 5 days a week. I mean, in Australia, they may be going once if they’re lucky. So how

David Trossell [00:09:50]:
do how do you see this

Karissa Breen [00:09:52]:
how do you see this unfolding then?

David Trossell [00:09:53]:
So okay. So if you’re working typically, what a lot of people now is put the the WFO and WF each people close to a cloud. In most clouds, you’re fairly close. And if you think about it, if you’re using an application in the cloud, you don’t really notice the, latency. You will do if you’re using an application in in the UK, but the application resides in New York. You will see the effects of it. You see the effect when you’re downloading a video as well. You know, the, the rotational dots that go around, that’s the latency coming into effect.

David Trossell [00:10:29]:
It’s just trying to pull that data across. So that’s how you, how you do it. So, in reality, latency is latency is latency. The only way around it is to use WAN acceleration, which is an intelligent way of filling the pipe up. So where that video that you’re trying to download comes across at a slow rate because of the latency, we can probably download it in a in a fraction of the second. Because what we’re doing is we’re filling that pipe up with data.

Karissa Breen [00:10:55]:
So in terms of other misconceptions people have around WAN acceleration, is there anything else that you can share as well? Because, again, you know, doing this podcast, interviewing people like yourself over the years, and then when they start sharing their thoughts and their insights about, oh, this thing that people think they know, this is sort of the reality of it. So I’m really keen to get into this with you because you’ve already sort of, you know, shared some interesting stuff so far, which I think people wouldn’t know.

David Trossell [00:11:21]:
I think that in one of the main causes of confusion in the network between ourselves and other products is that WAN optimization is all about reducing the amount of data you transfer across the network. So that compression algorithm that we that we use. And WAN acceleration is different. With WAN acceleration, we don’t change the data at all. So we don’t inspect the data. We don’t change the data. If it comes across in gobbledygook, it goes across in gobbledygook. And if you want to accelerate encrypted data, we don’t touch the data.

David Trossell [00:11:55]:
We just pass it straight through. So you’re getting a fast throughput through our boxes, out the other end. So that’s the difference. That’s the way that we can actually increase the performance across the the WAN by moving the lots of data in parallel streams, but still have a very low CPU and memory overhead. It makes us very, very light from that one. The other thing with WAN optimization is you store the data local as well as you ship it to the other side. We don’t store any data on our product. There’s no storage whatsoever.

David Trossell [00:12:27]:
So once that last byte has gone through our machine and now the other side, we don’t see any more of it. There’s nothing left in the box, shall we say, which makes, very useful in cybersecurity environments. What we can actually do there is if your environment is compromised, we can give you a, an ISO or a new box and you can load up your configuration from your PC, which we can, you can store on the PC and then start recovering. This is quick as that with all the other techniques. It takes you hours to rebuild the networks, rebuild the, deduplication tables, etcetera, etcetera. So that’s the biggest difference between us and WAN optimization. It’s a lot quicker. It’s a lot faster.

David Trossell [00:13:10]:
It’s a lot lighter.

Karissa Breen [00:13:11]:
And do you think people confuse or sort of think it’s the same between WAN acceleration and WAN optimization from what you’re saying?

David Trossell [00:13:18]:
Oh, very much so. When we first started talking to people across the US and across the UK, you know, these people, they the first thing they do is, well, okay, you’re just got another WAN optimization, but we tried to prove to them that we didn’t. We weren’t another WAN optimization. We were at WAN Acceleration that they stored in, they just said, well, no, that’s got the market sewn up. Interestingly now, Riverbed’s beginning to struggle with WAN Optimization out there. Or we are beginning to see large organizations and corporates encompassing our technology because it moves data so fast. And if you’re actually moving your data into a storage location for cybersecurity, the faster you can move it there, the less chance anyone has has a chance of intercepting it from that point of view.

Karissa Breen [00:14:04]:
Okay. So going back to Riverbed, why would you say they’re struggling? Is it because they’re on this optimization voyage rather than the acceleration?

David Trossell [00:14:12]:
Yes. That’s right. Because, thing with, we’re on optimization. You’ve got the local storage. So the more data you pass through it, the more storage you require, unless you unless you kill some out of it, you know, and call some of the data in there. But that’s, you know, it’s still a good technology for small offices. Have it and need to connect to large headquarters it’s still good. But then again, what’s happening with the cloud is people are putting their data in the cloud, so you don’t really need WAN optimization out there anymore.

David Trossell [00:14:44]:
But if you want to move all your data from one cloud to another, we can do that very, very fast.

Karissa Breen [00:14:49]:
So going back to the optimization, do you envision that this will just become obsolete in the in the future or not too distant future from what you’re saying?

David Trossell [00:14:58]:
It’s still encompass, or should we say embedded in a lot of SD WANs now to get try and give you that level of performance. But, with one of our customers who wanted to move data from Bangalore to Denver, which is about 230 milliseconds, which is a hell of a long way. You think about it In time, they had an SD WAN, and we’ve, bookended their SD WAN with our products. And that took the time that they estimated to transfer all the data from Bangalore to Denver from nearly a year down to a month because we can accelerate that SD WAN. So once the numbers of SD WAN suppliers get hold of this technology, we should see this embedded inside the product to give them that boost as well. Because again, it’s not that fast. You still got with SD WAN, you still got the lanes issues.

Karissa Breen [00:15:48]:
So you’re saying with going down the optimization route, it was forecasted to be a year versus acceleration, which was a month? Yes. That’s like 90% reduction.

David Trossell [00:16:00]:
Yes. People don’t believe it when we do it.

Karissa Breen [00:16:03]:
So then talk to me a little bit more. You said the term bookend. What do you mean by that when you said it?

David Trossell [00:16:07]:
Basically, we put one of our units between the source and the SD WAN. And at the other end, we put another one of our units between the SD WAN and the sync. So we’re actually accelerating the data across their SD WAN, and it works very, very well.

Karissa Breen [00:16:22]:
So going back to a year versus a month. So when you’re speaking to clients that you said before people don’t believe us, I’ve heard that as well from Tony, obviously, he works with you. How do you then start to prove, like, yes, we can do this within 30 days? Like, how do you how are you sort of demonstrating that that velocity?

David Trossell [00:16:41]:
It’s it’s difficult. So in most cases, we do a proof of concept because people just don’t believe what we’re telling them, and it’s still like it’s like, go away. You see, you’re in fantasy land. Not point of view. But if we if they’ve got a couple of machines, which, on VMware or Proxmox or something like that, which we can put a, a virtual instance on, we can connect them up and we can transfer data across the network. So we have a whole series of protocols. So let’s say you want FTP or Google or AWS, Amazon or any other cloud people, we can do that. So we could go in and out of the cloud or between clouds.

David Trossell [00:17:16]:
One of the tests that we did many years ago was a backup between Adelaide, I think it was, and south, West Virginia. So across the Amazon network, we actually did a backup for someone to prove them how much faster it was. Again, they were on phenomenal amount of latency. We took it down to somewhere like 90 times the expected performance. Same as, with a bank in the south, in South Africa, which is a big bank that had, banks also in London where they couldn’t get their GPR data out of their machines and into the bank in London in time not to be fined under GDPR rules. So we came along a little bit more difficult than normal, but because of their network, we managed to make their network work fast enough that they that they quickly were in GDPR compliance country very, very fast.

Karissa Breen [00:18:10]:
So then you said before the increase in their performance. So is there any sort of numbers or stats or insights that you can share that, you know, when you are speaking to these customers, like, for example, the the GDPR client that that one needs to move that quickly within a short amount of time, is there anything you can sort of talk through there?

David Trossell [00:18:29]:
Typically, what we would do is we’ll maximize their bandwidth. So whatever bandwidth they give to us, we will maximize it. But with one proviso that you can feed us fast enough with the data. And if we tend not to hit the maximum performance of the WAN link is because you’re not feeding us enough data. And that’s because it’s going through the machine so fast. And the AI is saying, come on, give me more, give me more. I’m running out of I’m running out of data to send, and that’s the way it works. But the other end, it will automatically de jitter, which means that if the packets come in slightly out of sequence, it will reorganize them to the back in sequence.

David Trossell [00:19:05]:
And it goes through. The whole thing is that it’s totally transparent to the protocol that you’re using to transfer across the data. You can have multiple protocols. In fact, there’s some customers that have different protocols. The other beauty about it is what we’ve done now for the cybersecurity side with our product. One acceleration is ability to turn off the connection through a calendar option. So that gives you that air gap, that very important air gap of between your headquarters and your backup site. And that’s key now for a lot of cyber security issues because where once they used to go after the live data, now they’re going going after the backup data first, corrupting that, then coming back and encrypting the live data, and then they ask ask for the ransom.

David Trossell [00:19:52]:
So there’s no way back for you. But if you can air gap that technology and air gap that backup, at least you got a safe copy that no one can get hold of. And that’s what we’ve been stored inside our machine now.

Karissa Breen [00:20:03]:
Okay. So I do wanna get into security. But before we do that, I’ve got a couple of other questions. You said before the AI is like, okay. You you gotta feed me more and more data. Like, how much data, like, is it wanna ingest? Like, how much are we talking here?

David Trossell [00:20:15]:
Well, it depends on the WAN. So if we have a 10 a 10 gig WAN, we’ll be looking to pull in around about 9.5, 10 gigabits of data. Don’t forget we got the what we’ve got there is a load of smart buffers in there. So intelligent buffers that one deals with the incoming data. One that manages the data into various chunks to optimize the way it runs across the network, and then there’s another buffer there that sends it out on the network. And at the other end, there’s the the corresponding buffers doing the opposite there. The idea is it manages the flow from the ingress to the egress. So it maximizes the performance all the time, and that’s where people see some phenomenal rates.

David Trossell [00:20:58]:
I think one of the very early ones that we did was in America, and that was between Phoenix and Rhode Island. And they wanted to transfer data between the 2 tape libraries, which they failed to do until we came along. We put our, our units in there and it took the backup down from 15, 15 to 18 hours down to 45 minutes. That’s the level of of acceleration we can give a customer.

Karissa Breen [00:21:23]:
So then I wanna talk about you said out of sequence. So you said it jitters and it’s out of sequence. How does that even come out of sequence?

David Trossell [00:21:30]:
Well, don’t forget what we’re doing with the data is marching into parallel streams. So the idea is, you know, the more parallel streams you have across the network, the more it fills the network up, which gives me gives you more data across the network. But some packets might go a different route, so they will come out a lot, come out of sequence. So what we will do there is because we’re handling so much data is we put these back in sequence. So they came into us in sequence. They might get lost and jittered as it goes across the network, but they the customer will have it back out in sequence as well. So the whole thing is totally transparent to the to the 2 servers, the the source server and the sync server.

Karissa Breen [00:22:08]:
Okay. So I have a question now about cost. So when people in the world think something’s faster, they automatically think it must cost more. Now I asked this question, and what comes to mind with the analogy I’m gonna use is when you go to a car wash and there’s, like, the express car wash, somehow that’s faster. It’s more expensive because they do it in 15 and not 45 minutes. You don’t have to wait around or, you know, for a couple of hours while they get around to it. So there is this sort of connotation with something being faster that’s gonna cost more. So would that would that be a fair assumption, would you say?

David Trossell [00:22:41]:
Yes and no. It’s one of those it’s certainly 2 on you know, double answer questions is, you know, lower end from our products where we’re very, very competitive up until 10 gigabits. Beyond there, we’re super competitive, but it requires them more performance out of your memory, more performance out of your CPU to transfer this data across. So as soon as we get up to 40 gig, we’re into more expensive controllers and CPU and memory systems. As we hit a 100 gig, we’re into a much into a, again, have a lot more memory, more CPU cycles. So there is a limit how far we can actually go from that point of view. How valuable is your data? That backup, if that’s your only backup you’ve got and the cyber cyber naughty people have corrupted to all your other backups, that one backup is very, very valuable to you. So that’s the difference.

David Trossell [00:23:33]:
You know, how valuable is your data?

Karissa Breen [00:23:35]:
Okay. So let’s now switch gears, and let’s flip over to security. So let’s talk about backups just for a moment. Would you say, David, in your career, there’s a lot of companies out there that don’t have I’ve heard don’t have any backups, maybe one or even that’s questionable. What’s your view then on that when it comes to backups?

David Trossell [00:23:53]:
The more, the better. There’s, you know, there’s there should be a 321 soon, but there’s 32 110 now. You know, it’s the, it’s the first call of port for the people who want to corruption your backup, and then you start asking for the ransom money. And people are still paying their ransom money. It’s still, it’s still amazing that people aren’t putting that amount of money into the backup systems because it’s once you’ve got an off-site golden copy, you can restore. You can restore your data. It might take your time, but you’re not gonna spend 1,000,000 of dollars paying someone to do it. And we know roughly from what part of the world they come from these days.

David Trossell [00:24:30]:
So backups are important. I spent most of my life in backups, working with tapes and all sorts of bits and pieces. And the thing that one of the things that most people fail to do is thinking about how you’re going to recover rather than how you’re going to back up because your backup process will determine your recovery process. And a lot of people don’t think that way. It’s people just put up, to put it out to the cloud or put it up to there. So there’s no guarantee you’re gonna get it back from the cloud. The cloud don’t guarantee your your money, your data will always be available. So it’s up to you to make sure your data is available.

David Trossell [00:25:06]:
Again, how quickly can you bring your company back up to strengths? That depends on your backup technology and whether you’ve actually optimized your backup technology that optimizes your recovery systems.

Karissa Breen [00:25:19]:
Okay. I wanna talk about you said before that people paying the ransom. So what I what perplexes me perhaps is, like you said, there are people out there that are paying these ransoms that that, you know, a lot of money, some of them, which could potentially, at times, exceed the cost of just doing the backup. So walk me through the logic here. Is it because people didn’t get around to it or we forgot about the backup or we don’t know what happened to the backup or the guy that was running the backups left 20 years ago, like I mean, this happens a lot to people. Like, I I I’m just trying to understand, like, as you put it in that way, it kinda makes sense that we don’t really need to pay it then because you got the backup.

David Trossell [00:25:55]:
How often do you do you miss paying up you know, paying installments for insurance? Backup has always been the Cinderella of any, IT systems, and people are beginning to put backups into the cloud. Again, know how secure is the cloud? Can you guarantee your data in the cloud? And if it’s in long term storage, how long is it before you get it back? And that’s the only thing you’re sitting around waiting. And we’re most people that pay those demand around some don’t get all their data back. Very rarely do they get all of the data back. So again, they’ve gotta re recreate some bits and pieces. You know, it’s typically you should have one side, you know, 2, you should have 2 backup sites, and they can be cloud. There should be 2 different cloud providers, so you can actually you know, one cloud provider goes down, the other one. I hear some people put in backups into a single cloud provider that’s got 2 halls, which is if the electricity goes down in both halls or you have a fire in both halls, you’ve lost your backup.

David Trossell [00:26:53]:
So again, it’s putting your data in disperse areas. Now we’ve done it over 5,000 miles. If you want to, we can do that, you know, and you can do that quite easily with our product. But also is, you know, from that point of view is having the data in in 2 different places. I mean, having it on 2 different pieces of media and having one on-site and one golden copy that’s perfect, which you’ve tested, knocked away an air gap so no one can get hold of it unless you unless you actually have the ability to get hold of that data. So that gives you that length, that depth of resilience on your backups. And you, you know, if you look at the articles that are going on now in various magazines is everyone is talking about backup because it’s the only way really. If they, you know, if you don’t if you don’t get the keys to unscramble your data, you’re stuck.

David Trossell [00:27:44]:
And that’s why it’s becoming more and more important. And it’s strange that tape again is beginning to come back again. You know, with LTO 9 LTO 9 tapes in huge capacity, you can get on those things now. Now you can go it’s safe because you can export it out of your library. You can put it somewhere else, air gap. No one’s gonna touch that. And that’s the that’s the beauty about tape backups. So I’m still a big fan of tape backups.

Karissa Breen [00:28:09]:
So going back to you said before, there there are companies that that have their backup in the same cloud. What what would be the rationale there? It’s just it’s cheaper to do that other than if you’re, you know, in 2 clouds, it’s gonna cost a lot.

David Trossell [00:28:20]:
It’s only a backup. Mhmm. You think yeah. You know, well, we’ll stick that as a backup. You know, here we do 3 copies. We do it, but we do a backup to a a cloud, a backup to tape and backup to tape, which we take offline. And that’s and we’re only a small little company from that point of view, the amount of data we have. I thought we got 3 copies.

David Trossell [00:28:38]:
And that’s what we you know, that’s important for us because it’s sort of our software costs it’s based across that lot. So even if you just do that for a small company, for a large company, it’s like, a large insurance company in America where they were having difficulties completing the backup. And it was this was a big IBM shop. Here’s that they were using dark fiber between the two data centers to send the backup data across, and they were never succeeding in getting all the data across. And we were pulled in to say, you know, can you help us? And we said, well, you’ve got dark fiber. What do what what do you mean? And we’re, no, we’re just not getting the performance. So we put our product into there, and we run a whole series of performance tests, which gave them a much better performance than raw dark fiber go because when we’re talking to storage area network devices like tape drives and disk drives, the signals you can do with the SCSI protocol to improve the performance above and beyond what we’re giving you with the parallelization. So it’s it’s quite a big step forward.

David Trossell [00:29:43]:
And before we finished all the performance testing, they put it into place because they could see the difference it made. What they were doing to get over this beforehand was doing a backup to flash copy to another disk array, then another flash copy to another disk array, and then they would try and back up the 3rd disk array across the site to the other data center. And that was never succeeding, so what they used to do is do a backup of that again, put all the tapes in the van and ship them across to the other data center and put them back into the tape library. So this is going on. And all the time, tape drives were underperforming, and that caused them to break, and they were breaking tapes and everything. But because we could use that dark fiber with our knowledge of the SCSI protocol on fiber channel, and then with our parallelization techniques, we could keep those drives spinning and streaming. And their tape usage went down phenomenally, and the amount of errors they had on their tape drives dropped lonely. So we had a secondary effect from that point of view of being able to keep their tape drives streaming nicely.

David Trossell [00:30:50]:
So that, again, that was in for that’s about about 11 years that was in there before they decided that we’re gonna go to the cloud, which that’s fine if you go. It’s interesting now that most of the cloud provided backup, their data to tape. So we just moved where the tape goes.

Karissa Breen [00:31:06]:
She said before, it’s almost like pendulum streaming back the other way. So you’re saying now more people are opting for for tapes?

David Trossell [00:31:12]:
Oh, yeah. The the the growth in tape is continuing.

Karissa Breen [00:31:15]:
It tape it off for a while when sort of the cloud really emerged in a little bit more.

David Trossell [00:31:19]:
You know? Yeah. Well, you know, we we used to do backups to tape to disks, which was a great idea until you looked at the reliability disk and you look at the reliability of tape. You know, tape reliability is phenomenal. There’s one, large bank manager who once told me, you know, once it’s out the once it’s out the tape, and I’ve been in store, no one can change it. That data was safe. And that’s that was a big, huge thing to me that, you know, I hadn’t thought about it that way. And that was why he did. He exported 500 tapes a day out of his big tape library.

David Trossell [00:31:52]:
It was it was huge. It was huge, but he said they’re the key tapes. They go into storage. No one touches them. That’s my recovery. With cloud, you’ve got you’ll make sure that the cloud provider’s got your data because they don’t guarantee it, and they don’t guarantee it’s noncorrupt as well. But if they, put it onto tape, then you’ve got that whole time basis for them to pull that back off a tape into your cloud instance and before you can put it back into your workshop, into your office.

Karissa Breen [00:32:20]:
So it’s fair to say that we will see a resurgence of the tape era?

David Trossell [00:32:25]:
Yeah. Even, you know, if you still look at it, my son worked in the tape industry in America. He was saying they were just selling tape libraries. I mean, so a lot of the, cloud companies are using tape libraries. I think I’ve heard that they’ve been signing their own tape libraries. I don’t know how true that is. But there’s people using tape. It’s it’s still the safest way of securing your data.

Karissa Breen [00:32:48]:
The only thing is that I work for a corporation maybe about 12 years ago. Someone came to collect the tapes. Couldn’t something happen, certain amount of tapes went missing, never to be found again. I mean, that’s obviously probably a rare thing, but I’m assuming that does happen.

David Trossell [00:33:00]:
Oh, I think one of our resellers who used to work with a tape encryption company, he went to New York and he found, let’s not say let’s not say who it was, but he said, yeah, how much is it to, for tape? He said, it’s $25 for an hour. So they would, you know, how many tapes do you want? You can read. And they would and some of the guy would set it out the back of the van, long as you sent it, long as you gave it back to him a little bit later. And that was a going rate, about $25 for a tape. So, yeah, there’s a big

Karissa Breen [00:33:30]:
leak there. So then how would you how would you recommend then securing that exactly? So someone come in, they do the tape collection, but then I don’t know. They have a car accident. Their whole car incinerates. We lose the tapes. They do exactly what you just said. They stop over and people are buying it.

David Trossell [00:33:44]:
So if you have the ability to make a local copy, and then there’s that same ability to make a remote copy without too much impact on performance, and that could be another data center. It could be another it could be another office or data center that you owned from that point of view. So if you own 2 data centers, even they’re 3,000 miles apart, we can still transport that data between that. So you can have a local copy and you can have remote copy. And the other data center can have a local copy and then send the remote copy back to you. So you can do cross site application. It’s easy.

Karissa Breen [00:34:16]:
So, David, do you have any closing comments or final thoughts you’d like to leave our audience with today?

David Trossell [00:34:21]:
The only way in in reality can recover from any cyber attack is having that golden backup, that air gap backup, that point of view. If you’re transporting data and it’s valuable, then encrypt it. We don’t want to encrypt the data for you. If you encrypt the data back in your systems, you keep the keys back in your system. So they’re away from us. They’re away from the way, and they’re away from the rest of it. Because we can accelerate encrypted data. To us, it’s all just ones and zeros, and that’s the best way of doing it.

David Trossell [00:34:53]:
That’s probably and then you can have that off-site capability, whether it’s in someone else’s data center or another data center of yours. That’s just that point of view.