RAID vs SHR - Why you should use Synology Hybrid RAID on your NAS

in this video I'm going to compare traditional raid implementations with shr or Synology hybrid raid and implementation based on the btrfs file system offered by a Synology on many of its Nas boxes if you want background on the various raid offerings how they work when to use them and how data protection works I have a video linked below but it isn't required viewing for this video Synology introduced shr in 2010 this now has extensive support in the home and SMB Nas range but they don't support this in their Enterprise class enclosures I'm going to come to why this is and how shr works at a deeper level in a follow-on video which I'm going to link below but the short version is that Synology don't seem to want to provide this as an Enterprise class feature to this point they also make the following disclaimer in the shrkb article implying that shr is not for advanced or experienced users however I am going to share why shr is actually very powerful and provides a lot of benefits over the standard raid options for many users as as you can see in this disclaimer they imply that shr is designed for people who are new to Ray technology and don't want to manage their raids but this is really only half the picture and we're going to get into that first of all how is shr similar to standard raid types and I want to make it clear that what happens under the hood is different but I'm going to talk first about the usability and functional similarities just like raid shr offers equivalent capabilities to the main raid types that people are likely to use for home and SMB use cases namely raid 1 raid 5 and raid 6. raid 1 provides a one plus one mirroring where for each disc containing data a second disk carries a copy that data ready to take over if a failure occurs shr provides an equivalent where all data has a copy stored in an additional disk to act as a mirror of that data raid 5 is based on data striping with parity using a minimum of three disks where parity is spread over those discs providing M plus one fault tolerance and allowing the failure of any one of these drives my RAID video explains how this works in detail but it basically means that if any one disk is lost the data exists to reconstruct its data from what is left on the remaining disks again f shr provides the same functionality spreading data over three plus disks and retaining parity information to reconstruct what was lost during a single disk failure raid 6 is similar to raid 5 but it provides n plus 2 resilience so that data can be recovered even in the event of no more than two simultaneous Drive failures terminology for these raid levels is slightly different with both raid 1 and raid 5 being called shr which provides fault tolerance for loss of one disk and shr2 providing raid 6 equivalence EG data protection for the loss of up to two disks so raid 1 for a single disc and raid 5 as an array are functionally equivalent to shr we've raised six being functionally equivalent to shr2 as I said what is going on under the covers is a little different between standard raid and shr and we will get onto that in a bit but the video I will link below will get you into deeper detail we then move on to what is different and this is where things get interesting as shr brings some really big benefits that make this a really compelling choice Beyond just Simplicity for non-advanced users with shr you can start with a one disk shr volume it brings no fault tolerance but you can then add a second Drive later to get M plus one tolerance in what is the factor via raid 1 mirror just like what is possible with Synology raid you can move from a mirrored pair to a stripe based array with one and then two disc fault tolerance you can also add a hot spare just like with raid with shr however as your array grows you can more easily mix the size of discs which isn't anywhere near as practical with raid this means you can grow the array based on need and buy the best value discs I would make sure however that as for any array you two discs of equivalent performance and reliability class so that array performance isn't impacted the easy way to do this is to buy the same model disks even if you change the sizes you go and of course stay away from SMR or shingle magnetic recording disks these should not be used in any Nas or in any disk array first of all I want to highlight a great tool Synology provides to help understand how the storage options grow when disks are added you can see it here and it helps you plan what to expect I will also link this in the comments below however it doesn't explain the underlying mechanisms and using this tool originally the logic behind it wasn't intuitive so I'm going to show you where these numbers come from and how to conceptualize how the data is laid out to compare how raid and shr treat disks and provide storage capacity and volume growth you can then easily understand the benefits of shr with traditional raid all discs are traditionally expected to be of the same size and usually of the same type you do not have to fill the enclosure on day one but you do need to select the correct raid type for example if you want raid 5 you need to start with at least three discs in this example we will say that they are 10 terabyte drives with this you get n plus one fault tolerance and the usable space of two of the three discs with the remaining 10 terabytes set aside for parity you can then add another 10 terabyte drive and grow the array to 30 terabytes available storage still with your 10 terabytes of parity and so on if you now choose to add a 14 terabyte disk you will only get 10 terabytes added as available storage and this is true of any disk that you add it has to be a minimum of 10 terabytes and no matter how much larger it will be treated as 10 terabytes of storage as we add 14 terabyte disks four terabyte of each is unused and if we start to add 16 terabyte disks this is even more pronounced once the enclosure is full the only way to get more available space is to upgrade all the disks to a larger size this needs to be done one at a time with the raid rebuild needed to complete each disc and it isn't until the final disc is upgraded and an array re-silver completed that the array will change to make use of the new disk sizes but again it will base the space available for each disk off the smallest disk in the array this means if one of your 10 terabyte disks needs replacing you could try and find a replacement disc that is the same as the failed unit which you may actually pay a cost per terabyte premium for or you can replace it with a larger disk but you don't get to benefit from the extra space until some possible time in the future when all your discs are replaced with larger units shr functions differently and it allows far more flexibility in your disk growth and allows you to get faster return on value from your disks let's look at the same example with shr as we looked at with raid we can start the volume with no fault tolerance and get 10 terabytes of usable space but with no data protection as we add our second disk we get no additional space but we do get our data protection in place now as we add the third disk we now move from a mirror to a stripe with parity adding 20 terabytes for usable space with 10 terabytes still set aside for data protection now we add the new larger 14 terabyte disk and at this point just like raid 5 we get just 10 terabytes added as usable space leaving 4 TB unused however adding a second 14 terabyte disk we start to see how shr releases more value from those larger disks it effectively creates a 5 disk array of 10 terabyte disks giving 40 terabyte but it also creates a mirrored set of the remaining four terabyte on the larger disks it then joins them together similar to a j-board to give 44 terabytes of usable space as we add our third 14 terabyte disk we now have in effect a six disc 10 terabyte raid 5 along with an additional three disc four terabyte raid 5 array with the remaining space giving 58 terabytes of usable space compared to the 15 terabytes we would have gotten from raid 5. as we had a 16 terabyte disk again just two terabytes is currently unusable but as the second 16 terabyte disk is added that third slice again becomes a mirrored pair creating a total of 88 terabytes of usable space with all the remaining 16 terabytes used for data protection again adding another 16 terabytes in this point would create a striped raid 5 setup in that last slice would you want an eight or nine disc or rain raid 5 is another matter but this is how it works so we can see here that not only can the array adapt to the number of disks with the method of data protection but it optimizes how quickly you can use storage from those larger disks if your first 10 terabyte disk starts to produce errors and you want to replace it you could just add a 16 terabyte disk in its place and get that as added usable space so you can constantly be growing with the current available best value for money disks and put the storage to use the way to think about this is that the smallest disk creates an array of this size and it uses an equivalent slice of all this to build a raid 5 array seen in the 10 terabyte box then if larger disks are available it takes the next slice and does the same thing once the array is providing data protection it needs at least two discs of this size to provide that this is highlighted in the four terabyte box and this continues as shown in the two terabyte box so each box is an instance of an array and then all of these are joined using jbub mechanisms to create one large representation which can then be used to create your disk volumes if you're running more than five discs I would also recommend using shr2 to give two discs a fork tolerance you can convert from shr to shr2 by adding a new disk and selecting to migrate to ssr2 rather than adding it to an existing volume shr2 is highly recommended as further errors during array 5 rebuild can occur and having two discs of Tolerance provides a cushion again don't neglect your backups as raid rebuilds can fail and your data can be damaged in other ways whether you use shr or Raid also note that if you're running shr you need two discs to get usable storage in each slice these are the 10 4 and 2 terabyte boxes I showed if you're running shr2 you need at least four in each slice to get usable space so conclusions and there are two major things to take away first of all shr provides a flexible data protection capability that allows you to grow your Nas storage as needed and cost effectively when your current array of disc is getting old and starting to suffer failures you do not have to replace the entire set of disks to get added capacity you can replace disks as you need and get storage growth as you go secondly there is the question of why Synology uses hedging language around it being for users who are new to raid or not giving support for its Enterprise Pro products I think this is some backside covering and to protect itself from issues in its larger Enterprise customers and because as I will cover in my next video on this topic it's based on a less mature technology however the technology appears to be robust and reliable and has been around for more than 10 years at this point in my following video I will go deeper into what is happening under the covers but the possible concerns here also really exist with the traditional raid option and going into this with your eyes open will help so three things you can consider raid is not a replacement for backup data can get damaged for lots of reasons only one of which is a radio failure you should backup data that you do not want to lose other than raid failure or data corruption the primary risk to any raid data is actually power loss and the causing partial data rights this can be mitigated by using a UPS and it's highly recommended that you protect your data this way even if the ups only provides protection for short power fluctuations or gives the nas enough time for a clean shutdown it can greatly limit the risk of data corruption and loss and in addition there is bit rot and this is where data gets damaged or degraded on storage while at rest cosmic rays unreported disk errors floating gate instability on SSD can all cause this and raid won't know it happen because data Integrity is only checked during the reading of data the way to mitigate this is to run a regular scrub and definitely run one of these when you get an unclean shutdown for reasons like power outages shr only runs on the btrfs file system and this provides a scrub capability process goes and reads all the data and checks it for consistency if it finds an error assuming it has a data protection copy it fixes the issue if you run this regularly say once a month you will greatly reduce your risk and scrubs are recommended for all raid implementations not just shr if you consider all of these and follow General best practices around backups that you should do anyway shr is a great option to give you the best flexibility in disk selection and allow you to get full value from these disks I run shr2 on my Synology Nazis and I have done for a while and they run reliably and I've had no issues doing days and scrubs or disreplacements so I hope this video was informative and I would really appreciate a like if it was also please consider a subscribe to get more from me I also welcome your comments below I know the topic of raid btrfs and ZFS can be topics people are passionate about and I love to hear your perspectives it helps me address interesting Topics in future videos and thank you for watching to the end and I will see you in the next

Channel: SomeTechGuy

Views: 24,309

Keywords: NAS, Synology, RAID, SHR, Synology Hybrid RAID, btrfs, ZFS, Disk Array, RAID 5, RAID 1, RAID 6, QNAP, Disk Station, RAID Parity, RAID Array, SMB NAS, Home NAS

Id: KuHXVnUPL0E

Length: 12min 44sec (764 seconds)

Published: Fri Jun 02 2023