RAID and Storage Solutions

in this video I will look at a number of different storage solutions these include basic solutions to more advanced enterprise solutions by the end of this video you will have a better understanding of what is available and how it can be used in your company in this video I first look at software versus hardware solutions next I will look at j-bot or just a bunch of disks and spanning both of these solutions allow multiple hard disks to be combined together to form the one drive next I will look at the more common raid implementations used in the workplace raid allows the administrator to combine multiple hard disks together for performance or redundancy reasons lastly I will look at storage solutions storage solutions allow the administrator to dynamically control how drives are allocated in this case I will have a quick look at Windows storage spaces and logical volume manager for Linux both these systems are excellent storage based systems and allow multiple drives to be combined and spaced dynamically allocated as required the first point to consider before choosing a software or hardware based storage solution is if you will use hardware or software a hardware solution will generally cost more than a software solution nowadays a lot of motherboards come with raid solutions at no extra cost if you start looking towards having a raid solution in a server with hot swappable drives this can push the cost of the server up for a server purchasing a system with raid and hot swap drives is recommended hardware has the advantage of taking some of the load off the system since some of the load is processed by the hardware this generally makes hardware faster than a software based solution a hardware solution will have the operating system seen multiple drives as one drive for this reason a hardware solution can be used to boot to the operating system in some cases you may need to add additional device drivers to the operating system in order for the OS to see the however it should be possible with a software based solution it may not be possible to boot the operating system depending on which configuration you choose this is because the software must first be loaded in order to access the drive in some operating systems you are limited to what you can choose for the boot drive it is a matter of checking the operating system and seeing what it supports for booting the advantage of software is that it may offer additional features as we will see there are software storage solutions that have advantages over hardware based solutions however enterprise level storage solutions may have these features as well but the cost of these features is considerably more than compared with the software the first solution that I will look at is j-bot or just a bunch of disks this solution combines multiple drives together to form one logical Drive unlike the other solutions that I will look at the drives do not need to be the same size the disadvantage of this system is that it offers no redundancy it only takes one drive to fail and all the data is lost j-bot also offers no performance increase the advantage of j-bot is that you can combine multiple drives together of different sizes speed and from different manufacturers but does not provide redundancy or speed improvements the next solution is called spanning typically a software solution spanning allows the free space on multiple drives to be combined into one drive let's consider the following example there are three hard disks with free space that is not being used spanning allows the free space from each of these drives to be combined into the one drive this allows you to use some free space that may have otherwise gone to waste however doing so offers no performance improvements or redundancy if any Drive used in the span set is lost all data is lost the next group of solutions that I will look at is Ray or redundant array of independent disks these solutions combine multiple drives together to form one drive the drives in the raid also known as raid array need to be the same size if you attempt to use a larger drive it will generally be allowed however the extra space on the drive will not be usable by the raid the drives in the raid should also be of similar specifications and speeds if you have a drive that is slower than the other drives depending on which raid you are using this can slow down the whole raid to the slowest drive depending on the features supported by the rate it may or may not support additional drives to be added to the rave later on this is the case with most rates in order to add additional drives to the raid all the data on the drives needs to be reorganized the more expensive rate solutions may have the ability to do this the cheaper options like raid included on the motherboard will not meaning that the raid will need to be destroyed and recreated if you want to expand the raid later on by adding additional drives I will now look at some of the more commonly used raids on the market the first raid level that I will look at is raid 0 raid 2-0 also known as striping is when two or more drives are combined together data is distributed across the drives in blocks as you can see when a file is written to a raid 0 Drive it is divided up into two parts and approximately half is stored on each drive as both drives can work independently of each other this improves read and write performance notice that as more files are added they are automatically divided up between the required drives in this example there are two drives however this could be expanded to an a number of drives for example if you had five drives using raid 0 potentially all five drives could be working at once to read and write at the same time you can see why raid 0 performance for reading and writing so good the downside of raid zero is that it has no redundancy if one drive was to fail you would lose all your data in this example if the second drive was to fail notice that the first drive only has approximately half the file information so effectively this drive is not that useful without the other one sometimes students have trouble remembering which raid levels have which features to remember raid 0 I like to remember that raid 0 offers zero redundancy the other raid levels that I will look at do offer redundancy in some form the next raid level is 1 raid 1 is also known as mirroring raid 1 requires 2 drives as you can see as files are written an identical copy is stored on each Drive making 2 copies in total since both drives hold the same information this effectively reduces the usable space by half even though this effectively doubles the cost of drive space it offers redundancy if the first drive was to fail the same exact information is available on the second drive it just becomes a simple matter of replacing the failed drive with another one and copying the data from the first drive the disadvantage with raid 1 is that there is no performance increase since both drives work together and have the same data on them reads and writes are effectively the same as using the one drive so raid 1 essentially doubles the cost of your space and provides redundancy but with no performance increase the next raid level is raid 5 also known as striped with parity this requires a minimum of three drives one of these drives however is lost to what is referred to as parity when data is written notice that all four drives are required to write information the last drive in this case contains parity information parity is additional information stored about the data a simple example of calculating parity would be to add the data from the previous drives and store the result on the last drive generally most r85 implementations will have the parity information rotated through each of the drives rather than store the parity information on the one drive using this parity information the raid drive can still operate if one drive were to fail or perhaps was stolen the drive can simply use the information on the other drives to work out what the data should be if a second drive were to fail all the data on the raid would be lost for this reason you should always replace the failed drive on a raid as soon as possible once the replacement drive is installed it will use the information from the other drives to rebuild itself raid 5 has additional advantages other than redundancy notice that when a file is read in this case three of the drives are used at once to read the file this means that read performance on raid 5 is excellent the more drives in raid 5 the faster the read performance notice however what happens when a file is written to the drive notice that all four drives are used including the parity drives since the parity also needs to be updated in some cases the other drives may need to be read to find out what data exists on the drive essentially this means that regardless of how little or how much data is being written all raid 5 drives need to be used this mix raid 5 very fast 4 reads but very slow for writes since raid 5 effectively only loses one drive for redundancy and offers performance increases for reading it has been popular in business but that popularity has reduced in recent years the main problem with raid 5 is if a drive fails it is possible to have a second fail before a new drive is and has had a chance to have the data copied onto it with drives getting larger and larger the rebuild time is also increasing and it is not uncommon in business to have raids with 10 or more drives in them if a second failure was to occur during the rebuild all the data in the raid would be lost so essentially there are two issues that need to be addressed with raid 5 the performance for writes and the ability to deal with multiple failures to address these issues raid 1+0 was developed sometimes referred to as raid ten raid levels that combine to raid levels together are referred to as nested raid the first raid level is raid 1 this works as you would expect rate 1 to work data is mirrored across the drives in the raid offering redundancy but no performance increase to offer a performance increase a second raid 1 array is added which is striped with the first using raid 0 you can see now that when data is written to the raid it is divided into two half the data is written on the left and the other half on the right this offers excellent read and write performance since the same copy of the data is held on two drives the rate can survive multiple failures the downside of this is that the cost to implement effectively doubles with the cost of storage decreasing raid 1+0 has become more popular in business is hard to tell how multiple failures are handled with only four drives so let's have a look at a raid 1+0 in a bigger implementation in this implementation of raid 1+0 there are eight drives in total half the available space is lost due to redundancy but as the data is spread across four drives read and write performance is quite good if a failure of one of the drives was to occur there is a drive with the same data available to recover from if a second drive was to fail notice that there is a drive next to it the same data as this drive so the rate is still recoverable if a third drive was to fail the rate is still recoverable the only time the rate is not recoverable is when a situation like this occurs let's say a DRI fails and then the drive on the second raid 1 array that contains the same data as the first drive also fails in this case two failures could cause all the data in the raid to be lost in this scenario there are only four sets of Mears drives however a typical business scenario may have a lot more drives in it imagine you had twenty drives with a total of ten raid 1 mirrors if this were to occur the chance of two drives holding the same data failing at the same time is quite low you can see that raid 1+0 offers redundancy but none of the performance problems associated with raid 5 there are also variations like raid 0+1 the difference is that the drives are arranged using raid 0 and then these drives are mirrored like raid 1+0 this offers the redundancy and performance improvements just like before and it can still operate after multiple Drive failures as long as two drives do not fail that have the same data on them the downside with raid and other solutions like these is they are generally not dynamic that is drives cannot be added and removed as required if you use a storage system this will allow drives to be added or removed as required they generally do this by placing drives into a pool if you for example had four drives in the one system of different types you can see here there's a mixture of solid-state and hard disks these drives could be combined into one storage pool the storage pool effectively manages the drives and their free space in the pool for the administrator giving the administrator the 1 pool of free space the administrator is then free to create drives using the space in the storage pool for example they could create a drive that is mirrored for redundancy in case a drive failed the administrator could also create a drive with parity for better performance or a drive with no performance or redundancy requirements if a drive in the system were to fail the administrator would need to replace this drive just like in a raid system once the drive had been replaced the storage system reorganizes the data to ensure that the drive meets the required specifications that is the mirrored drive has at least two drives with the same information and the parity drive is spread out amongst the available drives for better performance you can see that the storage pool handles all this for the administrator the administrator only needs to worry about replacing failed drives or installing larger or additional drives when more space is required for this reason storage pools are often used in large deployments where available space is often needed to be transferred between different customers there are a number of different storage systems available often in enterprise level san you will find features that allow storage pools to be created and managed in the case of this video I will look at two software based storage solutions the first sulfur-based storage solution that I will look at is Windows storage spaces this is a Microsoft storage system first introduced in Windows Server 2012 and Windows 8 it is aimed as a replacement for logical disk manager it allows the creation of drives inside the storage pool including the ability to have redundancy the redundancy is using mirroring or parity although mirroring traditionally uses two drives if you are using Windows Server 2012 r2 windows can use three drives to store the mirrored data on in the case of paradis the storage pool requires three drives if there are seven or more disks in the storage pool and you are running Windows Server 2012 r2 Windows can spread the parity out on the drives to ensure that the data can survive two Drive failures the next feature in Windows storage spaces is that it supports thin provisioning if you have any experience with virtual machines you have probably come across thin provisioning thin provisioning is when the drive only requires space that matches the size of the data as more data is added to the drive the drive becomes larger meaning it can dynamically expand as needed Windows storage spaces uses the new our EFS file system or EFS is designed to be a replacement to NTFS and offers additional features over NTFS at the time this video was created a drive created in Windows storage spaces cannot be used to boot the operating system this means Windows storage spaces can only be used for data at this time if you want your system drive to have redundancy as well you need to use a different solution like hardware RAID or software mirroring the next solution that I will look at is the logical volume manager or LVM this is a Linux based solution the LVM was developed before Windows storage spaces and has the same features logical volume manager allows for drives to be added or removed without any downtime drives can be created in the storage pool with or without redundancy the choice is up to the administrator once a volume is created it can be resized as required also the operating system can be booted from an l vm l vm does require the distribution to support it however most modern distributions do support l vm if you are using Linux it is worth the time looking into l vm this video has looked at some of the solutions available for storage for your computers in later videos I will have a look at how to implement some of these storage solutions I hope to see you in those videos and thanks for watching

Channel: itfreetraining

Views: 82,251

Keywords: Raid, Storage Solutions, ITFreeTraining

Id: 1h9sHu6jGmY

Length: 19min 44sec (1184 seconds)

Published: Tue Jun 17 2014