The Backup Wrap-Up

Ever had questions about SQL Server, Azure, SQL Server ON Azure, how to backup SQL Server, or how to backup Azure? This is the episode for you. Denny Cherry, a SQL Server and Azure specialist and author of seven books, talks to us about both of these technologies. Before talking about anything important, we tackle the mystery of how you pronounce Azure. Surprise! I was pronouncing it wrong, according to Denny, who talks to Microsoft people all the time.

We first talk about performance tuning, and Denny explains some things that most DBAs can do to improve performance, starting with indexes. (He also explains what an index is for those that don’t know.) We then talk about how bad query code needs to be in order to justify looking into that, and he gives us a few examples.

We also (of course) talk about backing up SQL Server, starting with the political discussion of WHO should own the backup process: a backup admin or a DBA? Denny and Curtis clearly do not agree on this one, but the discussion is a good one. Grab your popcorn! One of Denny’s best quotes is that he feels one of the primary jobs of the DBA is to be able to restore the database if something happens and if you can’t do that, nothing else matters. So beautiful.

Then the topic of dedupe comes up and things get heated again; our guest hates dedupe and Curtis loves it. That was another good discussion. Short version: make sure you have more than one copy of a deduped data store.

We continue the discussion of different ways to backup SQL Server, and Denny definitely prefers the native backup capabilities of SQL Server, and he explains why. Curtis then makes a suggestion on a way for DBAs and backup admins to both get what they want, but it doesn’t sound like Denny is taking the bait.

After a brief discussion on SQL Server vs Oracle, we move into the various ways one can use SQL Server in Azure. Denny’s gives advice as to what makes sense for most customers – and his opinion on the question of whether or not you save money in the cloud. Short answer: not usually, but you get a lot more power, flexibility and ease of use.

Regarding Azure vs AWS, it appears that Azure is very equivalent to AWS in overall functionality at this point, and there appears to be a number of cost and functionality advantages to running SQL Server in the cloud. One of the biggest advantages is that you can use an on-prem license of SQL Server in the PaaS version of it in the cloud. That’s pretty cool. We also talk about how roughly half of the VMs in Azure run Linux, and why that might be the case.

All-in-all it’s a really interesting podcast, even though we almost came to blows once or twice. (OK, not really.) But really good discussions about SQL Server, Azure, and backups of both.

Mark Lantz, Manager CloudFPGA and Tape Technologies for IBM, joins us on this week’s podcast to talk about how he feels that tape still has a future in data storage. We talk about past and future advancements in the substrates tape uses, as well as how tape has not approached the superparamagnetic limit, the way we have with disk. (This is the limit at which you cannot increase the storage capacity of a particular magnetic medium without creating more problems.) We have reached this limit on disk, where the magnetic grains have gotten so small, they can’t get any smaller without assistance. One such method of assistance is heat-assisted magnetic recording (HAMR), which we discuss – and how HAMR comes with its own problems. By contrast, tape hasn’t come even close to the superparamagnetic limit. In fact, tape can scale the aerial density 100X before it starts getting close. We also discuss coercivity and bit error rate (BER), which are extremely important concepts to understand. Another topic we talk about is how tape is getting better at scaling capacity faster than speed, because most people do not need faster tapes. (We talk about how and why we can’t stream the ones they have.) We finish out the podcast with an explanation of why helican scan drives (e.g. 8mm, 4mm, & AIT) all disappeared overnight. We cover a lot of territory in this episode, so buckle up!

Every wondered what you're supposed to do with all your old backups, now that you've moved on to another backup product? Simon Brown from StoneRam believes he has the answer to this problem that has plagued backup customers (and vendors) for ages. He's able to transfer backup data out of common backup formats and into your new product, or restore backups from your old product without having to maintain that infrastructure. It's a fascinating approach to this age-old problem. Check them out at https://www.stoneram.com.

After the recent stories about Veeam customers being directly targeted by the Conti ransomware group, we invited Rick Vanover and Dave Russell from Veeam to discuss the topic on the podcast. The stories in the press seemed to focus on the attack, as well as how ruthless the Conti ransomware gang tends to be. We thought we'd give Veeam a chance to explain exactly what Veeam customers can do to protect their backups from being exfiltrated and deleted. It seems that Rick, Dave, and company are doing everything they can to explain to all Veeam customers that this is something they should pay attention to. The following are two resources they said should prove useful:

Ransomware in 2022: 7 Capabilities You Need for Rapid and Reliable Recovery

https://bit.ly/3m32gI8

5 Ransomware Protection Best Practices

https://bit.ly/3nh7aAx

This week we are joined by Greg Edwards, CEO and founder of CryptoStopper, to discuss once again the important topic of ransomware. We talk about the challenges typically experienced by ransomware victims, especially exfiltration and potential exposure of sensitive data. The only way to stop that particular attack is stop the data from being exfiltrated in the first place. CryptoStopper has a way to detect that a ransomware attack has begun, but stopping it before it does any actual damage.

This one will get you talking! Jody Bruchon, author of “ZFS won’t save you: fancy filesystem fanatics need to get a clue about bit rot (and RAID-5),”, joins us on the podcast. The blog post went viral, resulting in three times as many words in the comments as the original article had. We start with an explanation of bit rot, why it happens, and why ZFS won’t be able to fix all bit rot. (For more information on bit rot, check out Episode 111 of this podcast here: https://www.backupcentral.com/tape-drive-designer-schools-mr-backup-on-tape-restore-it-all-podcast-111/). Jody then explains how ZFS needs disk redundancy in order for its self-healing features to work, and how if you don’t have that, you’re going to need backup to repair a ZFS volume damaged by bit rot. (We also talk about how it’s possible for a bit to be flipped without being noticed – even with ZFS.) Jody’s main concern is that people talk about how ZFS can be used to repair data corruption – without explaining how you need RAID-Z (or something) to use those features. He also explains why he prefers RAID-5 or RAID-10 to RAID-6. We then discuss “shucking,” the practice of buying external drives and ripping the drive out of them – to save money.

Curtis (Mr. Backup) then gets into an argument with Jody about the merits of Blu-Ray vs disk vs tape as a backup medium. Jody has some good points, but Curtis was unconvinced.

If you want to read Jody Bruchon’s original article, you can do so here: https://www.jodybruchon.com/2017/03/07/zfs-wont-save-you-fancy-filesystem-fanatics-need-to-get-a-clue-about-bit-rot-and-raid-5/

Inspired by the article "How good is your backup, really?" by Sandra Vogel, we discuss how to evaluate and potentially redesign your backup system. We talk about different kinds of backup systems, and how that impacts how you evaluate them. We also talk about how important it is these days to ensure that your backups are impervious to ransomware. We also talk about the importance of including recovery testing in your evaluation, and what kinds of restores to test.

Here's a link to the original article: https://www.itpro.com/server-storage/backup/357713/how-good-is-your-backup-really

We talk to Celeste Kinswood from Druva about cyber resilience, and specifically ransomware attacks that have long dwell times – which is most of them. The median dwell time (the time between infection and you finding out you have ransomware) is 23 days, and the average is around 90. It's encrypting files during that entire time, and responding to that is beyond difficult. Celeste talks about a novel approach that Druva is taking to solve this growing challenge.

We review and discuss Mark Dargin's Network World article "Credible threat: how to protect networks from ransomware." His article lays out several steps, each of which we discuss and expand upon: train your people, update your servers and apps, antivirus tools on endpointsm backup your data (of course), test your backups, and conduct vulnerability assessments (Pen tests).

Read the original article here: https://www.networkworld.com/article/3218708/how-to-protect-your-network-from-ransomware-attacks.html

We all know how important it is to verify your backups. But many companies simply lack the technical ability or time to do such a thing. Chris Marshall's company, VerifiedBackups.com to the rescue. They have a service that is aimed at companies with under 100 GB of SQL data. and will do an end-to-end verification of your SQL backups, guaranteeing that they are recoverable and safely stored offsite. I'm a fan of "as a service" anything, but this really takes the cake. Such an important thing to do, and he makes it happen.