The Backup Wrap-Up

We cover my latest book, Modern Data Protection, by O'Reilly & Associates on this podcast. I give an overview of the book, which covers the "Why, What, Where, and How" of backup, archive, and disaster recovery. After giving an overview of the book, I talk about why I wrote, and what it's like to write a book for O'Reilly. Prasanna acted as a tech editor on the book, so he offered his unique perspective as well.

Prasanna and Curtis discuss whether or not can (or will) OVH properly redesign their backup infrastructure to prevent another incident like what happened in March, where many customers lost their sites forever. As we discussed in our previous podcast, OVH had a backup service already that people paid for, and it was not up to the task. OVH"s CEO made an 8-minute video where he discussed some of the things they were going to do to make things better, and we discuss what he said. We talk about their idea of a centralized region just for backups, and whether or not that's a good idea. We also talk about how big of a job they have in front of them. We applaud what we see, but have many concerns that the brief video do not address. We also talk about how this plan is supposed to take five years, and what do OVH customers do in the meantime?

We talk to Mike Johnson of ComplyTrust, who says they "remember those you are supposed to forget." We talk about the data management challenges created by data subject access requests (DSARs), right-of-erasure (ROE, AKA right to be forgotten, or RTBF) requests, and the fact that we have many parts of the datacenter that are much better at remembering than forgetting. Backups are a particular challenge, but Mike brings up other challenges, such as mergers and acquisitions, and salespeople importing old data. ComplyTrust SaaS offering has a unique solution to this problem by remembering (on your behalf) those you are supposed to forget, and continually checking to see that they stay forgotten.

This week we discuss further lessons from the OVH fire, which starts with an admission by the CEO that some customers who paid for the backup service lost their backups in the fire. It then morphs into a discussion about designing resilient systems, starting with the concept of designing for failure. You have to protect against both physical and logical damage to your apps and data. We talk about using both cloud-native apps that have resiliency built in, vs having to add resiliency to your own app. Most importantly, know how your app/data is protected, and don't tolerate wishy-washy terminology in your service agreements. Above all, test, test, test!

This week we discuss a topic brought up by the OVH fire. It appears some people actually had a contractual backup service that wasn't doing it's job. How do you verify that a service you're paying for is real, and is doing what it claims to be doing? Especially how do you make sure they are storing data offsite? We've got some ideas.

Datacenter manager Dan Frith (@penguinpunk) joins us on the podcast for our first discussion of the #OVHFire. A massive fire destroyed a datacenter of a large cloud provider in Europe, and millions of websites disappeared. We talk about the lessons we can learn from this event.

Dan talks about how outsourcing the servers doesn't outsource the responsibility for data protection. I make the point that this fire shows what happens when you completely rely on a single entity for both production and data protection. We end up talking about the 3-2-1 rule and how it applies in this scenario.

I also give a discount code during the podcast for my new O'Reilly book Modern Data Protection, which is now available for purchase. If you use the URL below and the code I give on the podcast, you can get 35% of the retail price.

https://shop.aer.io/oreilly/p/Modern_Data_Protection_Ensuring_Recoverability_of_/9781492094050-9149

It only took us 100 episodes, but we finally got Dave Russell, VP of Enterprise Strategy at Veeam, as our guest on the podcast. Dave and Curtis go way back, and this was a great discussion. We cover the proper use of tape, and what it was like for Dave when he went to Veeam. Another big discussion point was Dave clearing up misconceptions (some of which may have come from this podcast) about what Insight Partners acquiring Veeam really meant. We then get into a great discussion about how Veeam works, ending that discussion with Dave explaining what Veeam is doing to address concerns about Windows and ransomware.

Our anonymous guest this week is from a Fortune 100 company who is considering swapping out their backup product. Our guest has been at the company for over 20 years, and remembers swapping out NetWorker for NetBackup many years ago. Now he is considering swapping out NetBackup to address his challenges with that product. We discuss a number of topics, including the age-old argument of who should be in charge of database backups, as well as the challenges of moving to a modern backup product when you are still using operating systems not usually supporting by such products. Our guest's final thoughts center on the importance of a good relationship with the vendor in question.

A Veeam user warns of what he felt is a confusing option in Veeam Backup for Microsoft 365. He says he likes the product, but that the first retention setting mentioned in the documentation (item-level retention) might not do what you think it does. He thinks everyone should use snapshot-level retention, which behaves more traditionally. We also discuss IBM Spectrum Protect (AKA TSM) a little bit, as they also use that product. Our guest is speaking on conditions of anonymity, so we gave him a fake name (Puddleglum) and altered his voice in the recording. (Want to talk about your environment, but don't want to use your name? We'd love to have you on and we'll keep you anonymous too!)

Adi Ruppin, founder of Ananda Networks, joins us on the podcast to discuss how they secure – and increase the performance of – network traffic without deploying a VPN or SD-WAN. He talks about how the technologies we use for networking are actually very old designs that come with a lot of downsides. Ananda Networks aims to address those downsides while giving you everything a VPN and SD-WAN do – and more. Faster and more secure internet connections without the technologies we usually use for such things.