Mr Backup spills the beans on 30 years in the storage industry

W Curtis Preston, aka “Mr Backup”, has seen a lot over the past three decades in the backup industry. 

Mr Backup’s LinkedIn profile encapsulates his career, starting out as a backup system admin for MBNA in 1993 through to becoming Chief Technical Evangelist at Druva. 

During those 30 years he’s written four books on backup, the most recent being Modern Data Protection published by O’Reilly in 2021. We reviewed it and said: “It is an excellent book, widespread in its coverage, detailed where it has to be, and using examples that we can all understand.” (You can download a no-charge eBook version here. )

Blocks & Files: You have 30 years in the backup industry. What have been the major changes you have seen in these three decades?

W Curtis Preston aka Mr Backup
W Curtis Preston

Preston: I can think of three seismic shifts in the backup industry. The first was commercial backup software starting to directly support the backup and recovery of applications, instead of just files. This became essential, as almost everything became an application – even servers (i.e. VMware). 

The move to deduplicated disk as the primary backup target would be the second big change. This solved the speed mismatch problem everyone had with tape. (Unlike what most people thought, the problem was that tape was too fast.) Disk is a much more versatile medium, even if it does cost more.

The shift we’re in the middle of now is the move to cloud-based backup & disaster recovery, and it was made possible by the two shifts I just mentioned. Everything we are backing up is now some kind of app; we are rarely just backing up some files on a server – and everything is being deduplicated in some way.

Blocks & Files: Do you think users are now better educated about why they should backup data and how backups should be organized in the 1-2-3 way?

Preston: I actually think we’re in a worse position. Sysadmins 30 years ago were running mission-critical servers on individual disk drives. No mirroring, no RAID arrays. And you would regularly lose disk drives, which would take a server with you. The same goes for laptops. This meant you were constantly reminded that you needed good backups, and constantly had a need to use them. Now anything mission-critical is RAID-protected, and most things are solid state. Today’s admins (and people with their own personal data) just don’t have the same respect for backups they once did, and assume far too much as a result.

Blocks & Files: What was the biggest advance in backup? Virtual tape libraries? Purpose-built backup appliances? Tape format consolidation onto LTO? The use of public cloud archives?

Preston: The biggest advancement in the last three decades – no question – is deduplication.  VTLs were a bandaid. PBBAs are being replaced by SaaS. Very few people are using LTO in their backup systems. Yes, many are moving their backups to the cloud. But virtually all advancements in backups in the last two decades have dedupe to thank for them, because dedupe makes the physics and cost of backing up to disk and the cloud possible. 

Blocks & Files: How would you evaluate the progress of deduplicating backup-to-disk technology?

Preston: PBBAs running target dedupe (deduping some other company’s backup format) were the predominant deployment method for dedupe in the first 15 or so years. In the last few years, though, a number of backup software and SaaS companies have figured out how to dedupe at the source – as part of the backup process. It’s where dedupe belongs because it’s the most efficient – if it’s done right. It saves CPU time on the client, bandwidth usage to the backup system, and enables backing up most datacenters either remotely to another datacenter or directly to the cloud.

Mr Backup book

Blocks & Files: Tape was the best archival storage back in the 1990s and it still is. Why is that?

Preston: It’s still cheaper than anything else. It’s 1-3 orders or magnitude better at writing 1s and 0s than disk (depending on which drives), and it reliably stores data 5-6 times longer than disk does. Object storage can deal with help with some of this, but it’s still going to be much more expensive. In fact, even if the disk were free, it would still cost more than tape over 10-20 years due to power consumption.

Blocks & Files: What are the main changes in tape autoloader and tape library technology during your time in the industry?

Preston: They just got bigger and more scalable. It’s much easier now to buy a moderately sized tape library that can grow to 10-20 times larger without wasting anything. They also added a lot of redundancy within the frame, meaning no single points of failure.

Blocks & Files: Ditto tape driver hardware?

Tape drives have gotten a LITTLE better at running at slower speeds without shoe-shining, but it’s still a major problem. The transfer speed (with compression) of an LTO-9 is almost a gigabyte a second.  No individual backup stream is going to come even close to that. 

Blocks & Files: How do you evaluate the progress of optical disk storage as a backup and archive medium compared to tape?

Preston: Optical has had a lot of fits, starts, and dead soldiers along the way. It’s a great medium for storing smaller amounts of data for very long periods of time, but it’s really never reached mainstream use in the datacenter. This is primarily due to speed. Optical units are simply much slower than disk or tape since the phase change necessary to write data that way is slow. I have my hopes out for one or two optical units, but they’re still on the fringe at this point.

Blocks & Files: When you started, SaaS backup wasn’t a thing and now it is. Why is that? Where is it going?

Preston: All of Druva’s competitors now have some kind of SaaS backup offering. Why? First, no one has ever wanted to do backup. It was always the job given to the most junior person.  Now it’s much more important than ever – and more important to do it right (and securely) than ever before.  You MIGHT get hit by a disaster, but probably won’t. But you sure as hell are going to get hit with ransomware. And every other ransomware story has part of it that says “and the backup system was also encrypted.” SaaS is a way to solve that problem for good by having your backups be set up and designed by people who do nothing but that. I see it becoming even more popular as people move more and more into the cloud. The more you are in the cloud, the more SaaS becomes a no-brainer.

Blocks & Files: Ditto running applications in the public cloud. How would you describe the realization that data in the public cloud needs protecting and the emergence of suppliers such as Spanning in the early days and Clumio latterly?

Preston: People understand they need to back up IaaS vendors, but most people don’t do it right.  They keep their backups in the same account and region as the original. They are just begging to be the next codespaces.com, who had their whole environment and their backups wiped out in a few keystrokes. 

As to SaaS vendors, the vast majority of customers think they don’t need to backup their SaaS vendor, like M365, Google Workspace, Salesforce. This is despite therebeing numerous examples of what happens when you don’t. They are simply burying their head in the sand and ignoring the fact that these vendors are not backing up their data.

Blocks & Files: Do you think it possible that any one data protection supplier can protect all environments: bare metal, virtual machines, containers, the public cloud, SaaS apps?

Preston: Most of those are possible in one vendor, with the exclusion of bare metal. That’s become a real rarity these days as everyone is assuming everyone is virtualized, and bare metal there just means backing up and restoring VMs. Having said that, there will always be new SaaS products that aren’t supported by vendors that try to do it all, giving rise to startups to support those apps.

Blocks & Files: How has ransomware affected the backup industry?

Preston: It’s suddenly made everyone pay attention to their DR plans. It has also created a much tighter beyond between the backup and security groups. Those of us who have specialized only in backup have had to learn a lot more about security. And it’s moved the security features of each vendor to the top of the “must haves” when looking at new products and services.

Blocks & Files: Do you think anything will challenge tape as an archive medium any time soon? DNA storage, for example? New optical storage tech such as that from Folio Photonics?

Preston: Optical is the only thing that could do it any time soon. If they make it more affordable and scalable. Folio Photonics looks like they may be on to something, but I haven’t seen it in the field yet.  Only time will tell. I’ve seen so many amazing products that seem to never get traction in the market.

Blocks & Files: What developments in backup and restore do you think should have happened but have not?

Preston: I think true immutability should have come sooner and stronger, and there should have been an agreement on what different types of immutability are. All backups should be able to be truly immutable, which means they’re not able to be deleted – even by a backup or system administrator. 

Blocks & Files: What has been the biggest failure of the backup industry?

Preston: The lack of protection it offered against ransomware attacks. Far too many customers were using recommended backup designs, only to find they were defenseless against a coordinated ransomware attack.

Blocks & Files: What has been its biggest success?

Preston: Modern day, cloud-based backup and DR that brings DR to the masses.  When I started, only the richest companies could afford to have a datacenter ready to spin in a few minutes. Now a company of almost any size can do that, and in a far more affordable way. Besides using the cloud to perform the recovery, modern day backup systems are able to do backup and DR with one system. For far too long backup systems did operational recovery, and replication-based systems did DR. Now you can do both with a single system and I think that’s great.