Two weeks before Christmas, my server died. I had all my data safely backed up – but on an expansion chassis that required a functioning server to access.
I confess, it took a long time for my Synology DS-1517+ to die; performance was slowing for months. It slowed so gradually that I didn’t really notice – except that copying files to the server took longer and longer. But, now, it was dead.
Sigh….
The only bright spot – if there was one – was that, hopefully, the drives containing the data were “probably” OK. It was just the server chassis that kicked the bucket.
After researching the Synology website, I bought their latest 5-drive desktop server: the DS-1522+. If everything went as planned, I would simply move the drives from the old unit, slide them into the new unit and, if the gods of data transfer were with me, everything would reset to normal.
The reality was not quite that simple, but after a few high-stress moments, all my data was safely recovered and I’m back in operation. Here’s what I learned.
EXECUTIVE SUMMARY
The Synology DS-1522+ is an excellent choice for a small office server. As I noted in my reviews from Nov. 2022, “For absolute speed, nothing beats storage directly connected to your computer. But nothing is better at sharing than a server. The Synology DS-1522+ is an excellent choice for any small workgroup.
“Fully Mac & Windows compatible, with vast configuration options, the DS-1522+ server delivers the maximum data transfer speeds your network will support. While speeds depend upon a number of factors, the capability of the server will not be a limitation.”
Upgrading to this system was a very smart decision. Server speeds more than doubled!
Manufacturer: Synology
Product: DS-1522+
Website: https://www.synology.com/en-global/products/DS1522+
Price: $699 from B&H Photo
Price is for an empty chassis. Adding drives increases the cost.
THE PROBLEM
Synology support is excellent – though not fast, answers can take a several hours, generally overnight. Their advice helped me figure out what the problem was and what was necessary to fix it.
All Synology servers come with a three-year warranty. If my server died within the warranty, they would simply have swapped it out for a new unit. But, of course, it didn’t. I bought the server about five years ago.
Worse, though, was that when I bought my first Synology server – with zero knowledge of how to configure a server and no money to hire someone to set it up – I chose the wrong file system.
Synology supports two file systems: BTRFS and EXT4. I originally picked EXT4 because I recognized the word “Linux” in its description. It took me several years to realize this choice was a mistake. In every regard, BTRFS is a superior file system, but, sadly, once a drive is formatted, the file system can’t be changed unless all data is removed and the drives are erased and re-initialized.
UPGRADING THE SERVER
Upgrading servers would normally take less than 15 minutes assuming the following conditions are true:
In this case, Synology makes upgrading server hardware fast and painless: just move the drives from the old unit to the new one. Most DS servers manufactured by Synology over the last several years support this disk transfer.
WHY BTRFS?
I will leave to others far more technical than me to describe the details of BTRFS. From my perspective there are three big advantages:
MY SYSTEM
This is my “server closet.” Yup, an actual closet. The DS-1522+ on the right, DX517 expansion on the left, and a NetGear 18-port 10G switch underneath. Firewalls and Internet modem are on a lower shelf (not shown).
The DS-1522+ holds five drives providing about 60 TB of total storage. I use the DX517 expansion unit to backup the main server. It has roughly 30 TB of total storage.
The Synology 10G Ethernet port add-on module.
Along with the server chassis, I also bought the Synology 10 Gbps Ethernet Module because my Mac Studio has a 10G Ethernet port.
NOTE: If your network supports 10G, the faster Ethernet makes a BIG difference in transfer speeds.
The dual-port 400 GB SSD caches using Synology M.2 SSD cards.)
I will probably also add one 400 GB M.2 SSD as a high-speed cache to speed transfers, but I haven’t bought it yet.
THE PROCESS
Once the server chassis arrived, the 10G port was attached and the drives installed – which took an hour – it was time to power up.
NOTE: There was some level of nervousness at this point, because I wasn’t really sure whether it was the older server or the drives that failed.
The server took about a minute to boot up, then two-three more minutes for the drives to come online – but come online they did. My life was no longer passing in front of my eyes.
Next, before doing anything else, I upgraded DSM 7 (their operating system) to the latest version and rebooted the server. This took 10-15 minutes.
If the drives were properly formatted, I’d be back in business. But they weren’t.
REFORMATTING
To reformat the drives from EXT4 to BTRFS requires copying all the data from the drives somewhere else, so they can be erased and reformatted. Then, copy the data back.
But I had 36 TB of data! I didn’t have any “spare” drives that were that big…. except, I did have the DX517 expansion unit, connected to the server via a SATA connection.
(Click to view larger image.)
Hyper Backup is a DSM utility that backs up data from the main server to an attached expansion unit or cloud storage. You specify which shared volume – or volumes – you want to back up, where the backups should be stored and how often you want them backed up, and Hyper Backup does the rest, including data verification.
My RAID has five volumes on it. The smallest is 1 TB, the largest is 30 TB. Using this utility I backed up all the data from the main server to the expansion unit. The backup process took roughly three days – about a terabyte every 90 minutes.
Once these backups were done, the second stress point arrived: Erasing all the drives in the server, reformatting them to BTRFS, then copying the data back.
Fortunately, the reformatting took less than ten minutes, but restoring the data from backups (my third stress point – was all that data safe?) again took several days.
THE RESULTS
(DS-1517+ measured by AJA System Test v16.2.5.2, using a 16 GB, 4K, 16-bit RGB file.)
This is the old DS-1517+ server – back when it was working properly and loaded with 36 TB of data. In practice, most of my file transfers averaged around 225 MB/second.
NOTE: 1G Ethernet limits transfer speeds to a maximum of 125 MB/second.
(DS-1522+ measured by AJA System Test v16.2.5.2, using a 16 GB, 4K, 16-bit RGB file.)
This is the new DS-1522+, again loaded with 36 TB of data. In practice, I regularly see file transfers at more then 500 MB/second. (If I install the SSD cache card, I would expect upload transfer speeds to approach the wire speed of 10G Ethernet.)
I am convinced this performance improvement is due to two factors:
SUMMARY
I am very happy with the Synology server, but, more importantly, that Synology’s software tools for data backup and restoration worked perfectly. While I lost a few days transferring files and had a bit more stress than I would like, my data was safe. Now that the process is over and I better understand file systems and backup software, I have all my data back – and double the system performance.
These are two excellent things!
4 Responses to Lessons Learned Migrating to a New Synology Server
Larry –
Thanks for this detailed story. I recently migrated from a Drobo 5D2 (which still works but for how long?) to a Synology DS-1522+, and have been pleased with the results. I used Carbon Copy Cloner to automate the process. I used Backblaze to back up the Drobo, but discovered that you have to pay much more to use Backblaze with the Synology server – bummer. As long as the Drobo works, I plan to use CCC to keep it in sync with the Synology server and use Backblaze to back up the Drobo. I will look for an affordable cloud service that can talk with Backblaze — any thoughts?
I also discovered FCPX Diet 2, which does a terrific job of removing render and other unnecessary files from FCPX libraries. I would be interested in your thoughts on Diet 2.
Many thanks for your always insightful news, tests, and commentary, and good luck with your new studio!
All the best,
Jan
Jan:
Thanks for your comment. (And, all comments are moderated, so they don’t appear the instant you submit them.)
I don’t use Cloud backups, so I don’t have a preferred vendor. But you might look into AWS (Amazon Web Services) or Microsoft Azure to see if they are cheaper than BackBlaze. I’m surprised they charge more based upon the storage device.)
As long as you’ve migrated your primary data to a newer server, using the Drobo as a backup in the interim.
I haven’t heard of Diet 2. I’m a fan of Final Cut Library Manager from ArcticWhiteness.com
Larry
Hi Larry. Great article as ever.
I recently had a Synology server die a few months after the 3-year warranty. The support team didn’t want to know, so I switched to a Qnap drive which is working perfectly. The process of transferring data took awhile using a Linux go-between.
The comment I have is a server is designed to run constantly in the back ground & most of the time if doesn’t do very much. Is it then ok for these expensive bits of hardware to fail so quickly? Should we not be expecting more reliability in their products? Or am I just too naive?
Sarv:
Thanks for writing. You have my sympathy on the effort it took to move your data to a QNAP server.
In answer to your question, older technology always fails. However, in most cases, I don’t think technology manufacturers are building gear with the intent of it failing. Would I expect a server to outlive its warranty? Yes. Servers are designed to run constantly, serving data 24/7 to multiple users. That being said, I generally expect computer products to last around seven years; at least for planning purposes. I have gear that is much older, but that’s a good benchmark to consider.
Larry