I recently installed a new Data Domain DD6300. Part of the whole installation procedure is to run a DD OS upgrade to bring the system up to the target DD OS release. You can find the target releases over here. While running the upgrade to 22.214.171.124, the Data Domain correctly rebooted as part of the upgrade. Logging back in, the system GUI kept throwing an “Upgrade in progress” popup, blocking everything else in view. There is also an alert that shows “DD OS Upgrade is in progress. The system will not be available for backup and restore operations. The alert will be cleared after the upgrade operation is complete.” Which I guess is NEVER when the upgrade is hung…
I’ve installed quite a few new Isilon clusters in 2019. All of them are generation 6 clusters (H400, H500, A200), using the very cool 4-nodes-in-a-chassis hardware. Commonality among all these systems is an 1GbE management port next to the two 10GbE ports. While Isilon uses in-band management, we typically use those UTP ports for management: SRS, HTTP, etc. We assign those interfaces to subnet0:pool0 and make it a static SmartConnect pool. This assigns one IP address to each interface; if you do it right, these should be sequential.
Recent addition to my install procedure is to create some DNS A-records for those management ports. This makes it a bit more human friendly to connect your browser or SSH client to a specific node. In line with the Isilon naming convention, I followed the -# suffix format. So if the cluster is called cluster01, node 1 is cluster01-1, node 2 is cluster01-2, etc. However, it turns out this messes up your SyncIQ replication behavior!
Roughly 6-7 years ago (around 2012), flash storage became affordable as a performance tier. At least, for the companies I was visiting. It was the typical “flash tier” story: buy 1-2% of flash capacity to speed everything up. All-flash storage systems were still far away into the future for them. They existed, and they were incredibly fast, but they also drove the €/GB price too far up, out of their reach.
However, in the background you could already hear the drums: it is going to be an all-flash future! Not just for performance, but also for capacity/archive storage. In fact, one of those people beating that drum was my colleague Rob. I recall vividly our “not yet!”-discussions…
And it makes sense. Solid-state drives are:
- More reliable: there are no moving parts in SSDs, and media failures are easier to correct with software/design.
- Power consumption is very low at rest: there is no little motor to keep platters spinning 24/7.
- Faster: the number of heads and the rotational speed of the platters limit a hard drive’s performance. Not so with flash!
They are still quite expensive, looking at €/TB. Fortunately, cost is coming down too. The last year or two, all flash arrays have taken flight in general-purpose workloads. Personally, I have not installed a traditional tiered SAN storage system in over a year anymore. Hyper-converged infrastructure: same story, all flash. The development of newer, cheaper types of QLC flash only helps close the gap in €/GB between HDD and SSD. But there is still a 20x gap. And one company we met at Storage Field Day 18 has a pretty solid plan to bridge that gap: VAST Data.
It is a fact of IT life: hardware becomes faster and more powerful with every new generation on the market. That absolutely applies to CPUs. A few weeks ago at Intel’s Data Centric Innovation Day in San Francisco, Intel presented their new Intel Xeon scalable processors. These beasts now scale up to 56 cores per socket, with up to 8 sockets per system/motherboard. This incredible amount of compute power enables applications to “do things”, whether it’s analytics, machine learning, or running cloud applications.
One thing in common across all applications is that they don’t want to wait for data. As soon as your %iowait is going up, you are wasting your precious and expensive compute power because the storage subsystem is not fast enough. Fortunately, WekaIO wants to make sure this will not be the case for your applications.
I recently expanded two 3-node Isilon X210 clusters with one additional X210 node each. The clusters were previously installed with OneFS 7.x, and upgraded to OneFS 126.96.36.199 somewhere late 2018. A local team racked and cabled the new Isilon nodes, after which I added them to the cluster remotely via the GUI. Talk about teamwork!
A brief time later the node actually showed up in the isi status command. As you can see in the picture to the right, something was off: the SSD storage didn’t show up as Isilon L3 cache. A quick check did show that the hardware configuration was consistent with the previous, existing nodes. The SmartPool settings/default policy was also set up correctly, with SSDs employed as L3 cache. Weird…
A few weeks ago I received a 1TB Western Digital Black SN750 M.2 SSD, boasting an impressive 3470 MB/s read speed on the packaging. I already had a SATA SSD installed in my gaming/photo editing PC. Nevertheless, those specs got me to pick up a screwdriver and install the new M.2 SSD. The physical installation is dead simple: remove graphics card, install M.2 SSD, reinstall graphics card. I wasn’t really looking forward to a full reinstallation of Windows 10 though. There’s just too many applications, settings and licenses on that system that I didn’t want to recreate or re-enter. Instead, I wanted to clone Windows 10 from SATA SSD to M.2 SSD.
After a little bit of research, I ended up with Macrium Reflect, which is freeware disk cloning software. Long story short: I cloned the old SSD to the M.2 SSD, rebooted from the M.2 SSD, and… was greeted with a variety of errors. The main recurring error was Inaccessible Boot Device, however in my troubleshooting attempts I saw many more errors.
The first SSDs in our storage arrays were advertised with 2500-3500 IOps per drive. Much quicker than spinning drives, looking at the recommended 140 IOps for a 10k SAS drive. But it was in fact still easy to overload a set of SSDs and reach its max throughput, especially when they were used in a (undersized) caching tier.
A year or so later, when you started adding more flash to a system, the collective “Oomph!” of the Flash drives would overload other components in the storage system. Systems were designed based on spinning media so with the suddenly faster media, busses and CPUs were hammered and couldn’t keep up.
Queue all sorts of creative ways to avoid this bottleneck: faster CPUs, upgrades from FC to multi-lane SAS. Or bigger architectural changes, such as offloading to IO caching cards in the servers themselves (e.g. Fusion-io cards), scale-out systems, etc.
A while ago I installed two new Isilon H400 clusters. With any IT infrastructure, consistency and predictability is key to a trouble-free experience in the years to come. Cables should be neatly installed, labeled and predictable. Wiring in the internal network cables, it helps if the nodes 1 through 4 are connected to switch ports 1 through 4 in order, instead of 1,4,2,3. While some might consider this OCD, it’s the attention to detail that makes later troubleshooting easier and faster. Like a colleague said: “If someone pays enough attention to the little details, I can rest assured that he definitely pays attention to the big, important things!”.
So I installed the cluster, configured it, then ran an isi status to verify everything. Imagine my delight when I saw this:
Several of our Data Domains are end-of-life and need to be replaced with new hardware. In most of the cases it’s a small site with a small Data Domain that only holds roughly 1 month of backups. In these cases we just install a new Data Domain next to it, reconfigure our our backup software, and that’s it. After a month, the old backups have expired and you can switch off the old Data Domain.
For the slightly larger sites, there’s more than one backup client/server writing to the Data Domain. There are Oracle RMAN backups, SQL dumps, etc. Plus the retention of backups on the Data Domain is much, much longer. In these cases you want to perform a proper Data Domain migration which retains the name and IP address of the old Data Domain, so you don’t have to touch all the clients. Here’s how you do that, and a DDBoost gotcha you should be aware of!
In my previous post I described how to reformat an Isilon node if for some reason the cluster creation is defective. After we got our new Gen 6 clusters up and running, we ran into another peculiar issue: the Isilon nodes lose network connectivity after a reboot. If we would then unplug the network cable and move it to a different port on the Isilon node, the network would come online again. Move the cable back to the original port: connectivity OK. Reboot the node: “no carrier” on the interface, and no connectivity.