Smarter, Strategic Thinking

Why Post-Production Studios Are Going Back to Tape

Fortuna Data Season 1 Episode 16

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 32:12

Send us Fan Mail

Ray Quattromini of Fortuna Data is joined by David Fox from JPY, the UK distributor for Archiware, for a deep dive into one of the most underappreciated areas of IT: data archiving, backup, and long-term storage management. 

David has been in the industry since the early 90s, starting out in pre-press and desktop publishing before migrating into the world of audio, video, and media storage. Today, he helps organisations across media & entertainment, law enforcement, universities, and enterprise manage and protect their data using Archiware's P5 software a powerful backup, archive, and replication platform that works across LTO tape, cloud, and a wide range of storage hardware. 

In this episode, they cover: - 

- The difference between backup and archive and why both matter 

- Hot vs cold storage and why media companies are constantly running out of space 

- How P5 handles LTO tape migration (LTO 8/9 to LTO 10) without losing your index - CCTV and law enforcement use cases, including 90-day retention compliance 

- Why LTO tape is making a comeback especially for AI training data 

- The real cost of cloud storage vs on-prem tape 

- Immutable backups and ransomware protection using S3 object storage 

- Data sovereignty concerns and the shift back to on-prem solutions 

- New features in Archiware P5 v8: EDL restore for video editors 

- How P5 is licensed and what that means for your infrastructure Whether you're in post-production, IT, or just trying to figure out where to put all your data this episode is packed with practical insight. 

Find out more about Archiware: www.archiware.com 

Find out more about Fortuna Data: www.fortunadata.com

SPEAKER_01

He's got his podcast energy.

SPEAKER_02

Yeah, got my shirt on.

SPEAKER_01

Yeah, don't wobble everything. Oh, that's fine. I won't um wobbling's fine.

SPEAKER_02

You can wobble if you want, though. Don't worry about it.

SPEAKER_01

Yeah, you do it. You do it. You're amazing, Ray. You're amazing. Right, okay. Let's let's try again. Okay. Let's go with your amazing.

SPEAKER_00

Your creative work is irreplaceable. But is it protected? We're sitting down with the experts who solved the storage nightmares keeping post-production teams up at night. From LTO tape to cloud archiving. This is a conversation you can't afford to miss. Let's dive in.

SPEAKER_02

Hi, I'm Ray Quatramini from Fortuna Data. Welcome to another edition of Smarter Strategic Thinking. Today we had David Fox from JPY, who are the archyware district. Hello, Ray. Nice to be here. Thank you very much. How long have you been in the IT industry? Because you look like you have been.

SPEAKER_01

Well, I have always been in IT since I did a computer science degree in the early 90s. So I've I've worked for PC resellers putting i386s in schools. My dad worked for a local newspaper as the kind of artist that would use rotary pens and rulers and things to build adverts and things and graphics to go in the newspaper. And when I was a teenager, they just dumped a load of Macs in his office. And they had like half a day's training or something like that from a local Apple Centre. But they spent a lot of money on kit and a laser writer as well. My dad said, Come in have a look at this stuff. You like computers, because I was into home computers. So and I absolutely fell in love with all of it because it was like the best kind of creative tools that were available in the mid 80s to late 80s. So I've always wanted to use Macs, always did when I was younger. I briefly had a job where it was just PCs, I didn't like that. So I got a job working for the company, the parent company of the company that I work for now, basically. We sold software for creatives in the late 90s, was more kind of pre-press and publishing static stuff, page layout, quite express. And we sold software that let all of those Macs link to a central server. We sold pre-press software into pre-press systems. But with ArcuWare, that kind of morphed into more of a creatives were more doing moving stuff. So audio and video. Because the the prepress guys didn't the file sizes that were at the time big and needed specialist kit to handle like printing photographs and stuff like that. That became easy, and then the harder stuff became the video stuff, and that was so that's where the more interesting storage storage challenges went when ArcuWare sort of came into its own.

SPEAKER_02

So do you want to explain what to our audience here what ArQare is and what you do?

SPEAKER_01

I work for the UK distributor for ArcuWare, but also I do a fair bit of sales and marketing directly for ArcuWare, who are a software vendor in Munich, so it's a German software company. Relatively small, they're sort of 20, 30 people. And they have a product called P5, so ArQWare P5 is a backup archive and replication tool. What P5 does in the creative space is that it lets those have a much better data management to protect the assets that are on their storage. So typically these sorts of media customers are either either producing post-production VFX for TV and movies, let's say. There's a certain amount of audio going on as well. So they have storage that they're using to edit from, where they've got workstations where they're putting together all of the cuts and so on, and they have lots of high-res footage on the server. So the server has to be fast, quite large, because it's got to handle media files, always filling up because as soon as a project's finished, another one's coming in, and that job's done, but they can't delete it, but they have to keep it forever, but it's quite big. So they have an archive need whereby they need to push stuff into what we call generically cold storage, right? So hot storage is the spinning disks and the SSD cache and everything for for editing. Cold storage is where they push stuff to. So that's the archive workflow. The cold storage that we support is LTO tape and cloud, where LTO is more popular. You can also use that same storage for backup because you need to protect that storage in case you know the building burns down or the thing stops working, so you can recover it from a backup. So two separate things. At the same time, you're archiving because you're filling up that storage all the time and you've got projects coming in and going out. So you need a workflow where you can move stuff to cold storage. So those customers are using cold storage in two different ways. And maybe doing a cloud backup as well, because if you're doing if you're backing up, you should you should also think about having it going to two different places. But you can do that with tape as well. You can take tapes off site and have two different uh backups in different locations. So yeah, P5, archive and backup. Obviously, P5, maybe not obviously, but P5 is just a software product. So it needs to be combined with the hardware, the LTO tape, the cloud subscription, and we work with all of the LTO drive and uh library manufacturers. So we're entirely agnostic with regards to the uh the hardware that we work to. We're not tied to just working with quantum or IBM or whatever, which makes us a nice thing in the channel. And then the final piece of the P5 software is called synchronize, which replicates data between hosts, but it's entirely cross-platform. So P5 runs on Windows, Linux, Mac OS, FreeBSD. It's in the store for QNAP Synology, NetGin, as is. And then we have uh the software pre-installed on a lot of proprietary storage boxes because we run in a Linux OS, and most storage is tends to be like an embedded Linux OS of some description. So we can run directly on those operating systems if we're allowed to by the vendor. Or you can put us inside of Docker, which means that we're kind of more portable, more easy to install where we're not allowed to run directly on the operating system. Yeah, backup archive and replication via the synchronized tool. And synchronise is good because it can synchronize between a QNAP and a Synology, or a Mac OS and a Windows, or any any combination of those operating systems, and it can now also synchronise to cloud um buckets as well. So you can just take a folder on your storage on-prem and just replicate that to a cloud bucket and then have that update each day to compare the, see what's changed and so on. So the so synchronise is very mixture of platforms and types of storage for having just a simple file and folder-based replication, which is sometimes all people need, you know. They they just want a second copy somewhere else rather than the complexity of a backup or or an archive.

SPEAKER_02

We, as you know, we did a test on our QNAP that's sitting over there. I'm looking at that. We loaded your software on, we did a local backup to the hard disks and the flash in the in the QNAP, and then we replicated it to Impossible Cloud.

SPEAKER_01

And it and it worked flawlessly. And it was the first time I'd I'd seen Impossible Cloud, and we were able to just, you know, create a demo account, yeah, get those credentials, plumb them into P5's um S3 connector, and we're off to the races.

SPEAKER_02

It's very quick and easy. It's a really good solution because obviously with a QNAP we can put in a SAS card or a fiber card and connect it to a tape library as well. So it made it, you know, your your solution supports it out of the box, so it made perfect sense for us to test that.

SPEAKER_01

Any LTO drive from any vendor, any library, any size, any number of drives in the library, we support all of it.

SPEAKER_02

So uh so you do eight, nine, ten, for example. Oh yeah, we go all the way back to one. If you go from LTO9 to LTO ten, you can handle that migration through your software as well, is that right?

SPEAKER_01

Yeah, so within the archive piece, we have a thing called Data Mover. In fact, it's been renamed, so it's not called data mover, it's called Data Life Cycle Management. If you've been archiving to a set of LTO L LTO9 and LTO8 tapes, and you've got 50 tapes that you've already got archive data on, and all of that archive data will be indexed within P5, so we'll know about every file on those tapes. We can often even show you a thumbnail of that file and metadata and so on. So what we do is we we go through the index or a subset of it, we find each file on the LTO8 and 9 tapes, and we just write it to the LTO 10 drive and then point the index at the LTO 10 version of that file. So the index remains the same. The index has still got your you know 1.5 million uh media files sitting in there, but now they're pointing to LTO 10 tapes instead of LTO.

SPEAKER_02

So if we've got an LTO 8 tape, for example, and we point it to an LTO 10 tape, are we saying theoretically we could get three LTO8 tapes on one LTO 10 tape?

SPEAKER_01

Oh yeah, yeah.

SPEAKER_02

So you So you're not it's not a one-to-one.

SPEAKER_01

No, no, no. Yeah, so we'll fill up the bigger tape. So you'll have you'll be reading lots of small tapes and writing one big tape. Right. So that's the beauty of it, because uh, you know, you don't need to have so many tapes. But what we also do is we rather than moving the data, we can copy it so that both copies could still be restored. So rather than having all of those eight and nine tapes and just throwing them away and saying we've migrated it, you really want to keep them because it's another insurance policy, isn't it? Put them somewhere. Exactly. And P5 still knows where all the data is on those tapes, but day to day you'll be restoring from the 10 tapes, and new stuff will go on the 10 tapes.

SPEAKER_02

Obviously, in the world today, I think some of our viewers will know this is uh it's very hard to buy any type of storage at the moment, whether it's a disk drive or a flash drive or an MVME drive, it's it's gone crazy. One of the things I know you mentioned that you know post-production and media production, but one of the other markets is obviously day-to-day businesses that have you know storage systems and they need to look at how they're gonna free up space on their primary storage because guess what? We can't buy it for a year.

SPEAKER_01

You work on Windows servers? Yeah, we work on Windows servers. You can install us, connect up a tape library, let's say. So let's say you buy a two-drive LTO 10, LTO9, 25 slot library. You can browse your storage on the Windows server, or you know, on what broader infrastructure, so you can install us as an agent on various bits of storage, and then you can archive centrally through the server that's got the hardware, the tape hardware attached. You browse your storage, pick a folder, archive it, and then sit there and watch the job in the job monitor window and watch it streaming to the tape.

SPEAKER_02

And could could we, if we're doing an archive job, and obviously you should always back up, you know, archive to two or more devices, yeah. Can we do a simultaneous archive to two separate tapes? You can, yeah. So we haven't got to copy it and then copy it again.

SPEAKER_01

No, so that's why people buy libraries of two drives. So you can run one job writing two tapes, both of them are full, take one out, drive it somewhere different, and then keep one in the library for restores. And if if the one in the library wears out or doesn't work anymore because maybe you were restoring from it every day. Because you put something on there that you shouldn't really have archived, you should have left it on the storage, then uh you can recreate you can regenerate uh the the the bad tape from the good one that you took off site. So we've we've we've got a workflow for regenerating tapes. So you can regenerate bad tapes, uh, you can write them in pairs for for redundancy. You can um you can migrate when you decide to get the next drive and the bigger the bigger version, um, when the price points are correct or whatever, because the drives are all you know they work in the same libraries. Um yeah. And you can also verify. So if you've got tapes that you wrote 20 years ago and you haven't migrated them yet, you can get P5 just to put them in the drive and just scan through and see if they're readable.

SPEAKER_02

And one of the things I read the other day is P5 also supports LTFS.

SPEAKER_01

It does for the archive, yeah. So all of the things that we talked about, the tapes can be written in the LTFS format. Pre pre-LTFS, RQWare had their own tape format, which is also the one that we use for backup. So backup we use the the native tape format. But for archive, yep, LTFS is fine. And then those tapes could be read by something other than RQWare. Do you do you want to explain what to our viewers what LTFS is? So LTFS linear tape file system, I think it's since version five. Five. But somebody said the other day four. No, it's fine. But I don't believe that other person's got it wrong. So LTO5, 6, 7, 8, 9, 10, linear tape file system. Um invented by IBM. Invented by IBM, and there's a software stack that IBM is used by third-party products, not P5, we do everything ourselves. But basically it's a dr like it's like a driver that allows the tape to be presented as a disk. Because the it was it was thought that the ideal way to access tape is just plug it in and there's an F drive or a T drive for tape, right? And then you just start throwing files on there, but brilliantly in that way. So presenting a tape as a disk, because a tape is much slower, uh not slow, but has a latency in order to get to the right position. It doesn't behave like people expect this to behave, so it upsets people if you try and uh use a tape in that way. But the the the important thing about linear tape file system is that it's it's vendor neutral. It's open standard, isn't it? It's an open standard. So if you've got an LTFS tape full of important data 20 years from now, there will be software that can read that tape regardless of the Arcuare, regardless of whether ArcuWare are around. Which of course they will be, but maybe I won't be, so I won't care.

SPEAKER_02

Retiring soon, are you? Well, I'm only 35. I view LTFS, it's just like one big USB key. Yeah. With with 30 terabytes or whatever the take.

SPEAKER_01

30 terabyte USB stick that you can actually fit in your pocket.

SPEAKER_02

From from your perspective, your market place, do you sell to you know, you you talked about post-production and media. Are there are they corporate customers and government organisations and institutions that you deal with?

SPEAKER_01

The majority of customers are media and entertainment. We've also, I mean, most recently we've put some licenses into the big universities because they particularly like the IT and the science departments, they create lots of data. There are some incumbent LTO tools that we've replaced in those cases because they were already using LTO tape and they already had tape libraries and they already knew all about the tech, but they just liked our software because it was simple. We've got some licenses with police surveillance. It's a bit like media and entertainment. They wear body cams, you know, CCTV cameras, all of those surveillance and law enforcement areas, they they create loads of media files. They don't really know what to do with them, as with I mean, the ME people are quite good because they've had this problem for a long, long time. As the resolutions go up for CCTV and it's 4K, now they have a problem with storage filling up. And uh, yeah, they at this point in time they just can't keep adding arrays. The the ability to to move that stuff off is is vital. And the the workflow with P5, the default workflow is you can browse a folder, put it on tape, now it's archived, then you could delete it or mark it for deletion later. That folder is now visible through our web interface. You can browse folder structures with files, so you can find stuff that you want to restore, you can search for file names, we can keep metadata and you can search that. You've also got like a visual search because you'll see a grid of thumbnails when you're looking at some photographs or videos. And then you choose stuff that you want to restore, and then we'll pick the tapes and restore it back to disk. So the workflow, the round trip is just browsing, choosing folders and saying go to tape or come come back from tape. There's an API if uh if a third-party tool like a MAM Media Asset Management or even CCTV systems, they want to talk to us using you know commands, they can do that and they can put stuff with us and take it back out again.

SPEAKER_02

A few years ago, now we used to sell a lot of CCTV systems, and then with the advent of the cloud, people started to say, can we store it to the cloud? you know, and that's what they did. But the problem is, as you quite rightly say, is when you get higher resolution cameras and you get maybe a hundred or two hundred cameras, putting that resolution and that data in the cloud becomes a bottleneck problem from a bandwidth point of view.

SPEAKER_01

Yeah, and you spend all your time using all of your internet bandwidth to try to do the a backup that's been running for a week and you're not quite sure when it's going to finish.

SPEAKER_02

When you're trying to do a playback from the cloud, it's not exactly fast. Hence the reason you have a local storage device so you can play it back. One of the things, even from a CCTV perspective, is they have to keep the data, I think some of them had to keep their data up to 90 days, then they have to delete it. From a CCTV point of view, is we could load our keyword onto a server, attach a tape library, and we could we could move that CCTV footage onto a tape library every 30 days or whatever, and then after a period, we can then delete it. You can delete it, yeah.

SPEAKER_01

And our archive uh workflow allows you to delete folders from the index of stuff that you've archived, and then we'll tell you which tapes are now empty, so that then when you've got an empty tape, you can then reuse it, re-reformat it, relabel it, and and add it as a fresh tape, or destroy it, and that way you can fly with that keep for only keep for 90 days.

SPEAKER_02

Some of these CCTV systems going back 10 years were like two three hundred terabytes then. They probably need more than half a petabyte now, and they need to keep it as I say for 90 days. But if you can't buy any storage of any description, then how do you change your workflow? Because people are still walking around shopping centres and town centres, you know, airports and everything else.

SPEAKER_01

And you can get completely stuck, yeah.

SPEAKER_02

And then all of a sudden you find, oh, we can't expand anymore, we've got another 200 cameras coming online. And I I know some systems that we were talking about have a thousand plus cameras. So it's quite interesting the market and how your product would help alleviate that solution. AI, when you ingest LLMs and you get your output, where are we going to store that output?

SPEAKER_01

And it's a it's a process, isn't it, where they have a huge amount of data that's needed in the training and then they're moving on to a model, they build the model, but the training data they don't need that, and so that could go to tape. The the cheap, like Glacier Deep Archive, AWS storage, and I think Google and Azure have the same thing.

SPEAKER_03

Yes.

SPEAKER_01

It's all tape. It is all tape. Cloud. So it's just a question of do you pay Amazon to put stuff on tape for you? Tiny little pipe, or do you do you just you you know not outsource tape, but buy some tape hardware and some tapes, and then effectively have limitless capacity just by buying tapes, which aren't scarce.

SPEAKER_02

LTO 10 tapes are quite expensive compared to LTO9 tapes. They're still available, and LTO 10 E tapes are horrendously expensive. 40 terabyte on an LTO tapes quite is very good. Why I look at it is yes, they are expensive, but there are a lot of benefits to LTO 10 because you don't have to format the tapes and wait an hour for the tape to format and everything else. So there are you know it's instantaneous.

SPEAKER_01

There are benefits and the tape and the tape cost is going to come down. We're not sure how quickly it'll come down because it's quite a bit more than LTO9. Yes, you know, it's future-proofing yourself by buying the latest.

SPEAKER_02

But the what the way I look at it is if this is an archive, if it costs you 300 pounds for something to put on a shelf, put forward to that that movie, that film, that data, whatever could be £100,000 or a million pounds, or so that's where I think the cost of storage, and sorry to pee off a lot of people watching this, but the cost of storage I think came down to such a low price, it lost its value. It's it's reasonably easy, we can just buy it off the shelf, we can get more of this, we can get more of this. And now prices have gone up three, four, five hundred percent in the last twelve months alone, rising, and lead times are six, twelve, eighteen months. It makes people think that £300 for £40 isn't really that much for something that could last 30 years.

SPEAKER_01

It'd be really a lot a lot of money when you compare the cost of an LTO tape with the cost of two or three LTO9s. Because if you're comparing it with the cost of disk storage from a year ago or now, it's still cheap, cost per terabyte. And it's not just the fact that it's cheap storage, but it's like the the nature of tape, half-inch tape on a reel in a box, is that it's very simple, and when it's not actually being written or read, it's just a box with a tape in it. So it'll last, you know, 30 years, is I think what the manufacturers say, which is, you know, you probably struggle to have a drive 30 years that'll read that particular tape gen, but it's very stable, which is something that hard drives aren't.

SPEAKER_02

Well, even we we had Kingston in our previous video, they even said organizations were looking to keep their data on uh NVMe and Flash, but over time NVMe Flash degrades, and after 10 years, you've got a blank NVMe and a flash drive.

SPEAKER_01

And it's a bit of an unknown, isn't it? I plugged in an old one terabyte Samsung like SATA drive that I'd just put with a cable with a US and it worked fine. But I was it was years and years and years since I plugged it in, and I was kind of surprised and kind of delighted that everything was still on there. It wasn't stuff I needed, but I was just curious, you know. But you it's a bit of an unknown. You don't quite know what the point is where that's going to stop working. I mean I've found like memory sticks in drawers from like 10 years and and they've still worked, but I'm always curious to know if they still work. Whereas like a reel of tape, unless it's been exposed, unless you got it wet or something crazy happened, it's it's a very simple piece of story. Tape has that advantage, doesn't it, as well? And also the portability. So if you've got 10 LTO tapes and you want to get them to a different location, you can do that quite easily. Whereas you can't pull drives out of RAID arrays and and put them in different locations. You've got to buy a RAID at another location and then replicate data to it. Let's be honest, tape's dead. But it is dead, obviously, yeah, tape's dead.

SPEAKER_02

How many clients and servers can Archieware actually protect?

SPEAKER_01

The way that it works is a sort of work group infrastructure. So you have a host with with the tape. If we're going to talk about tape, you have a host with tape hardware attached because the tape hardware isn't a network resource on its own like a NAS. It needs To be driven by some software on a host. So you've got a host and the tape together. So you can then install P5 on any other hosts where there is data that you want to secure. So, you know, maybe you've got four NAS devices and you've got two NAS devices at another location. Then you would install P5 on all of those and you would add them to the central P5 server. P5 server would know about its local machine, but also it would know about these four NASIS and these other NASIS at this other location. They'd all be listed. And then you can just browse that storage within our web interface. So we handle the file transfer over the network, and then you can archive and restore willy-nilly between these different hosts using our web interface. The only thing you have to think about, tape needs to, you know this, tape needs to be written and streamed at a reasonably quick data rate. Even a hundred mega second is a bit slow for a tape, but a hundred mega second is a how fast you'll get data over a one gig Ethernet connection. If you've got a NAS with a one gig connection to the P5 server where the tape is, the data's going to be coming along at sort of 70, 80, 90 megabytes a second. And that means that the tape isn't able to keep moving. The tape has to keep stopping and buffering and and and starting. And that's bad for every all of the tape pieces. It's bad for the tape, which will get stretched over time. All the mechanics in the drive will have to do more. You know, 10 gig Ethernet's fine, but one gig Ethernet not so fine. And if it's site-to-site links, then maybe you don't even have a gig. So you might want to copy that data first to a disk nearer to the tape.

SPEAKER_02

Which makes sense. You know, some of the systems that we sell from Seagate, we we create a local copy and then we replicate it to the other site, and then we can back that data up on the other site because disk to disk transfers and also you know TCPIP, and we get some data loss and you know latency and all these other sorts of things come into play that all affects how your backup's going to end up over here, isn't it?

SPEAKER_01

Yeah, exactly. And you you can always put in a separate Ethernet network just for that transfer between the boxes so that you're not competing with users accessing the storage. That's quite a good thing to do as well.

SPEAKER_02

But in particular, in the last 12 months, companies are starting to re-evaluate where their data sits. Because if it's in the cloud, that's all well and good, and it might be affordable and it's fantastic. But then you've got data sovereignty, US Cloud Act coming in. So there's a lot of parameters that organizations have never considered, but in the last 12-18 months are starting to come on people's radar again.

SPEAKER_01

The thing, the thing about immutability reminds me, I was talking to one of our resellers last week, and they're currently using P5 to synchronize a bunch of uh storage within the publishing company to another server that's like the backup, and they're just synchronizing files and folders. But the the management have come and said this needs to be immutable so that we are covered for our ransomware insurance. Yes. And so what they're going to do is either empty that storage and represent it as S3 with immutability as an option. So you access the storage as object storage with immutability, and then P5 could write to that as an S3 bucket.

SPEAKER_03

Right.

SPEAKER_01

Or completely replace it with some storage appliance that is S3 by nature.

SPEAKER_02

That QDAP over there supports immutability. You can set it up for immutability as well. Or or S3.

SPEAKER_01

You have quite a few choices with that box. So if you're backing up to it with P5, what we're writing to that immutable storage is just objects, chunks of data, not individual files, but chunks of compressed data. And those files, we never try to delete them, we just add files as the backup goes along.

SPEAKER_02

With the companies that back up to immutable storage in the cloud, they make an API call to AWS, and for every every block of data that they send using the API call, AWS charged the customer.

SPEAKER_01

API call.

SPEAKER_02

So is that the same with ArcuWare?

SPEAKER_01

Yeah, if you if we're writing to a cloud bucket, then it is. Then you you have to pay the bill, and the bill's comprised of how much did you store, for how long, how many API calls did you make to just look at what you've got rather than and then upload and download transfer fees. And then there's minimum retention as well. So if you're using Glacier, there's a is it six month, 180-day minimum retention? So if you put anything in a Glacier bucket, even if you remove it the next day, you'll pay for six months. Wow.

SPEAKER_02

What trends are overhyped right now and what are what people are not paying attention to, in your opinion, from a point of view of storage? Where do you see the storage market panning out in the next couple of years?

SPEAKER_01

I think I see a lot of resellers that are people that are using cloud-based storage in media and entertainment for editing. Um, products that basically present a file space via a client that you run on your workstation from anywhere, and then they they pull the files and they work really well and they allow for distributed workflows, but they're all the eggs in one basket solutions because if they're you've got to trust 100% that they look after your data. So we're seeing quite a few of those users that want to have a on-prem copy of everything, sometimes to tape, sometimes to disk. Tape more attractive going forwards. Yeah, we're working with a few different vendors to do archiving to an on-prem tape library. So where you could have a tape library here in this room, but you could present that as an S3 bucket on a public URL, just like an Amazon S3 URL, and then P5 can receive data via the S3 connection, cache it for a bit on disk, and then dump it to tape and then empty the cache. And then if that software comes back and says, Oh, give me that object, we'll pull it back to disk and then and present and provide it back. So very much like Glacier, in that it's it's an S3 bucket, but it's got tape behind it, but where you host it.

SPEAKER_02

Yeah, well, as I say, that our QNAP supports S3 storage, supports S3 tape, you know, and it's that's it. I think the next interface, you know, I know one one of the we were we were an IBM Roadshow as a partner last week, and one of the things that IBM are talking about is you know, we we've traditionally had fiber channel networks, we've had SaaS connectivity, we've had Ethernet connectivity. But IBM are seeing a huge uptake at the moment of NVMeOF, so NVMe over fiber because the speed of some of IBM's flash systems are like 80 gigaseconds.

SPEAKER_01

Yeah.

SPEAKER_02

And when you're creating AI models, you need to ingest and dump rapidly.

SPEAKER_01

So it's a bit like going back to SAN. Yeah. Because everything moved from fiber-based SAN direct block access from clients to storage. That became network-based, didn't it? Ethernet based and NAS network attached storage. But you're talking about the storage being so fast, again, eye watering, right? That that you have to access it over fiber.

SPEAKER_02

Eye watering. Some of the some of the systems that they're talking about to some of their huge banking institutions, they're running at 600 600 terabits a second.

SPEAKER_01

600 terabits a second. Yeah. It's unimaginably fast. Which it is. But when you've got like these server farms doing AI training.

SPEAKER_02

So look at Nvidia's connections and things like that. It's all optical. You know, some years ago I saw a thing called Silicon Photonics, and Silicon Photonics is terabit networks at microsecond transfer rates. This thing is coming down the tracks rapidly. What's next on the roadmap for Archieware?

SPEAKER_01

Recently we've added some features that are kind of specific for editors. So going back to Media and EDL restore. So edit decision list. The the editing job is basically I've got a hundred shots from a camera. I'm going to use my editing software, and you have a timeline, and you start putting the clips on the timeline, and then you build your movie or your TV program. And that's a very skilled job, you know, the program is made in the edit, so this software is very sophisticated. What an EDL is, it's a way of exporting the timeline as a text file. And there's a few different formats, but it basically describes the timeline, the project. And so in it, it has all of the clips that we used and the bits of the clips that we used, and that can be used by ArcuWare to restore uh the original media files from tape. So if you um if you if you shoot all your 4k, 10k, whatever video, you would probably generate proxies because the the original files are too big to edit with at speed. So you generate bunch of proxies, you put all of that footage on tape, you archive it off, you don't need to store it on your disk storage. You do the you edit the project, you come up with the final timeline, and at the end there's a thing called conform, which is where you need all of those original files, the ones that you've used, to come back to make the high-res perfect copy, and then you might do colouring and stuff like that. So we can we can sit in there and we can take the EDL file, like and we we can just restore the files that we used. Because like half of them, 80% of them probably won't use, didn't make it into the edit at all. So we introduced that a couple of months back in version eight. That's some a feature that everybody gets. So that's that that's kind of the direction we've been moving in a little bit is specific stuff for editors to make just just to refine what we've got.

SPEAKER_02

So how how is your software licensed?

SPEAKER_01

It's it's licensed to a to an extent around the size of the tape hardware. So number of LTO drives, number of slots. If you've just got one drive, you just need to license one tabletop drive. If you've got an 80 slot library with two drives, then you need to license 80 slots and two drives. So we have sort of baseline products that come in with two drives and 25 slots. And then if you add another 25 unit on top, you can add another 25 slot license. So, yeah, broadly that's that's how it works. It's if if you add more LTO drives or slots, then you'll need additional licensing. But the base stuff is is all covered. Right, Dave. Nice talking to you, Ryan.

unknown

Good.