Use examples of BaaS¶
Use Case №1¶
User story: A backup regime is required for an instance (virtual server) provided by SIM-CLOUD using the BaaS service.
How should this be done, how to begin, and what requires special attention?
Solution: First, it is necessary to determine the objects for creating backups. The BaaS system was designed to create backups of the virtual disks of an instance. Depending on the architectural solution, an instance may contain multiple disks and for each disk the system allows an individual algorithm for creating backups.
It should be realised that the system operates in fact with backup objects broken down by disk, and not with the entire instance overall. Such a mechanism allows the backups to be managed more flexibly. All or individual disks of an instance can be created or restored; disk space can be saved by creating or omitting backups only for defined disks.
After the objects are defined, i.e. the disks for creating backups, it is necessary to narrow down the number of restore points; this is the number that will define the ‘depth’ of the backups. The availability of a large number of restore points will allow the disk to be restored with older data, which may be needed, depending on various requirements and circumstances. It is also important to bear in mind that the fundamental resource of the BaaS service is disk space, which is restricted by the disk quota purchased by the client. The volume of backup objects overall, together with the number of restore points, will determine the overall amount of disk space required for proper system operation. A shortage of disk space may result in the process of creating new restore points being interrupted, and as a result, the loss of the possibility of restoring current information in the event of a malfunction.
Having understood the principal aspects described above, we may now consider the direct configuration of the production of backups. For this it is necessary to create an appropriate task that will be launched by the system in accordance with the settings chosen. When creating the tasks in the dialog box, the following must be specified: name of task, its description, and the required object: the disk that has to be copied. In the task scheduler the time and number of restore points should be specified. After the data are completed, the system launches a backup in accordance with the time defined at the configuration stage.
The backup system allows two approaches to the creation of tasks to be used: the first is to create a separate task for each disk, each of which is individual in terms of its settings for launch time and backup depth. This solution is appropriate if the project does not involve too many instances and as a result does not include large numbers of disks that have to be backed up. The second approach is more systemic and is suitable for users who require backups for large numbers of disks. Such a method of organising backup tasks is scalable and thereby reduces the time required for backing up. This method is based on the creation in advance of tasks with sound logic for creating backups. By creating the tasks the user systematises the needs for backing up particular disks, and in so doing typifies the tasks. Once a number of tasks have been created, the user needs only to add the required disks and the BaaS backup system will perform the backup operations in accordance with the scenario initially prepared. An example could be a situation in which 5 instances are created in the cloud, each of which has a bootable system disk. For each of the disks the backup scenario is the same, as a result of which it is possible to create a single task, give it a comprehensible, intuitive name (system disk backup), select one system disk, set up a schedule and specify the number of restore points. After the task is created, the user can add/remove similar disks to the task in no particular order, and it also becomes possible to correct a task, thereby allowing changes to be made for the whole group of disks and tasks.
Users of the BaaS backup system should bear in mind that the backup object, i.e. the disk, can be added only to a single task (the same disk cannot be used in two or more different tasks). Once the disk is added to one or another backup task, the disk will not be available for the creation of other tasks. Should the latter be necessary, the disk should be deleted from the existing task and a new task created on its basis.
Use Case №2¶
User story: The system does not allow the deletion of one of the disks from an instance and generates the following error message: ‘Error: You are not authorised to perform the operation: delete disk.’
Reason: This response by the system may be because the disk is set up on the system for BaaS backups and on the basis of which snapshots are created. Attempts to remove a disk that is used for backup purposes from the ‘Disks’ section will result in the above error occurring.
Check whether this disk is linked to the backup task. If it is, that disk cannot be deleted without first changing its configuration. To delete such a disk, first disconnect or delete it from the corresponding backup task; this will delete the administrative snapshot. After this, ‘uncouple’ the disk. To do this:
- Go to the ‘Backups’ section,
- Select the backup created for the disk,
- From the drop-down list in the ‘Actions’ column, select ‘Disconnect disk’.
Once these actions are completed, the disk can be deleted.
Note that after completing the tasks above, the disk for which the operations were performed will no longer create backups via BaaS. If backup is required, this will have to be set up again from scratch. ______________________________________________________________________________
Use Case №3¶
Disk recovery does not work because there is not sufficient free space in the project storage.
Users of SIM-Cloud and its backup service BaaS use two independent storage spaces, one of which (the primary storage) is intended for storing the volumes of the instances, while the other storage is for backups of these volumes.
Recovery of a volume from backup is to the primary project storage. During the recovery process, BaaS does not replace the existing volume in the main volume storage; rather, it performs a recovery in the storage area specified. For this reason it is necessary, before recovering a volume, to ensure that there is sufficient space in the storage to which the recovery will be made.
For the user, the primary cloud storage consists of the quota of available disk space. The primary storage is created on the basis of two disk drive types: gs1 and cs1.
By default, recovery of a volume from backup is to the same storage type as that on which the volume was created. If there is insufficient quota for recovery on one of the two storage types (cs1 or gs1), the volume can be restored on the other storage (where there is sufficient free space), and subsequently it will be possible to launch an instance from this disk.
Users of the BaaS service can choose the type of storage to which the recovery is made (cs1 or gs1).
As an alternative, the existing volume of the instance can be deleted and a recovery made from the backup. Using the space freed up from the deleted volume, it is possible to restore the volume with the required information on a sequential basis in accordance with the available restore points. When deleting a disk it is important to realise that it will also no longer be possible to restore the data in its previous configuration. Only the existing points (state) of the disk at the moment that the backup is performed will be available for recovery.
The optimal and recommended solution is to increase the quota of primary storage (cs1 or gs1). With this approach it will not be necessary to delete the volume for which a recovery is performed. _______________________________________________________________________________
Use Case №4¶
According to the BaaS system, our available disk space has run out, which is the reason for errors when trying to create new restore points.
BaaS disk storage is limited by the quota that was purchased by the client. Over time, the free disk space may run out and it will then not be possible to create new restore points. The reason for this may be an increase in the number of disks to be backed up, or the number of restore points. Perhaps the volume of used space on these disks has grown, resulting in increased backup volume. The reason could also be more intensive data exchange on the disk, which also affects the space occupied by restore points in the storage.
The following approaches are available to resolve the situation:
- The user can purchase additional space for backup.
- The user must reexamine their backup policy. For example, reducing the backup depth will reduce the number of backup copies accordingly and thus also the space occupied by them. When the tasks are next launched, first a rotation occurs of the existing restore points, and then a further check of space is performed and the backup process begins. This means that if 5 restore points are configured for the task, changing this figure to 3 means that the system will have to launch 3 times. After the third restore point has been created, the backup rotation is completed and the system performs a re-count of the available disk space.
- Optimisation of backup tasks. An example of this would be to consider creating two tasks, one of which will back up the most important disks and store copies for 10 days, while the other backs up less important disks for just 3 days.
Exceeding the disk quota will prevent backup copies being made, which may result in it not being possible to restore current information in the event of a failure.
Use Case №5¶
It is necessary to restore a disk from a backup. How should this be done, how to begin, and what requires special attention?
To restore a disk, go to the ‘BaaS backup’ menu and select the ‘Backups’ section. In this section a list of available backup copies is displayed by disk. Find the required disk and restore point, based on the creation date, and click the ‘Restore’ button. After a period of time the disk is recovered and can be found in the ‘Disks’ section of the ‘Computing Resources’ menu. If it necessary to restore a disk, it will be possible to connect it to an existing instance or to create a new instance based on it. When restoring a disk, take care to ensure that there is sufficient project disk space available.
Use Case №6¶
It is necessary to completely restore an instance from backup. How should this be done, how to begin, and what requires special attention?
It is best to break this practical case down into three parts:
- creating the instance;
- recovering the required instance disks;
- connecting the recovered disks to the new instance.
When performing a complete recovery of an instance from backup because it has been deleted, note that the data will be updated to the moment of the creation of the restore point. If an instance has been deleted from the project, the link to the network settings will also have been deleted. The system issues network configurations (IP addresses) dynamically. When an instance is deleted, previously assigned IP addresses are freed up and may be given to a different client or a different instance. When a system disk is restored, the operating system will use the former network settings, and specifically the same IP address (network mask, gateway and dns). For this reason, a new IP address should be issued in the OS network settings that was assigned to the instance when this was created.
Use Case №7¶
How do I define the required volume of the data storage quota for creating backups?
The size of the disk quota for the backup system can be estimated from the following considerations:
- Minimum disk quota size is 10 GB.
- The disk quota should be a multiple of 10 GB.
- The volume of storage for BaaS is estimated as the total used space of each disk plus the sum of changes (delta) of a disk between restore points:
VBaaS = (VHDD1 + (∆VHDD1+…+∆ VHDD1n)) + …+ (VHDDm + (∆VHDDm1+…+∆ VHDDmn)) n = 1…k, where k-is the number of restore points;; m =1…f, where f-is the number of disks to be backed up.
From this it follows that the required disk space is affected by:
- the number of disks in the project to be backed up;
- the volume of used disk space to be backed up;
- the number of restore points;
- the intensity of data exchange on the disk between restore points.
Three disks have been created for the project, for which backup is required. It is necessary to estimate the required disk quota for BaaS that must be purchased.
The details of the initial situation can be summarised in a table:
|1||Volume of disk space||120GB||250GB||500GB|
|3||Volume of used disk space (VHDDm)||30GB||100GB||350GB|
|3||Number of restore points («k»)||7||5||10|
|4||Intensity of data exchange on the disk between restore points (∆VHDDm.)||1GB/d||5GB/d||30GB/d|
The formula given above can be used to estimate the required disk quota:
VBaaS = (VHDD1 + (∆VHDD1+…+∆ VHDD1n)) + …+ (VHDDm + (∆VHDDm1+…+∆ VHDDmn))
After analysing the initial data it can be concluded that in the case of constant intensity of data exchange on the disk between restore points (∆VHDDm), the formula can be simplified to the following:
VBaaS = (VHDD1 + (∆VHDD1 x k HDD1)) + (VHDD2 + (∆VHDD2 x k HDD2)) + (VHDD3 + (∆VHDD3 x k HDD3))
Substituting the initial values into the formula:
VBaaS = (30GB + (1GB x 7)) + (100GB + (5GB x 5) + (350GB + (30GB x 10)) VBaaS = (30GB + 7GB) + (100GB + 25GB) + (350GB + 300GB) VBaaS = 37GB + 125GB + 650GB = 812GB = 820GB
As can be seen, the result from the formula (812 GB) is then rounded upwards to the next divisible by 10 GB, i.e. 820 GB. The estimate shown enables an optimal evaluation of the BaaS disk quota required for configuring backups. To calculate the maximum quota, the entire disk volume is taken into account together with the maximum number of restore points and the maximum intensity of data exchange between restore points.
A shortage of disk space may result in the process of creating new restore points being interrupted, and as a result, the loss of the possibility of restoring current information in the event of a malfunction. It is therefore necessary to approach the process of selecting the backup disk quota carefully and also to choose the optimal configuration of tasks for backing up the disks.
Use Case №8¶
It is necessary to restore a 500 GB disk and to determine more accurately the efficiency of the recovery process and the time required for restoring the disk from the backup. Does the content of the disk affect the speed of recovery?
The architectural solution implemented in the BaaS service allows retrieval of some 200-300 GB per hour. There is no difference here in terms of the types of data processed by the system. In essence a recovery point, a slice of the client’s system status and the file system are created with data blocks. Since a backup copy is created on the basis of a disk and a defined instance, the recovery from the backup also takes place at disk level. For this reason, SIM-Cloud BaaS can be applied equally effectively with all types of infrastructure services, be they a web server, a database server, a mail server or an ordinary home operating system. The content of the disk does not affect the restore speed.
The following formula can be used to calculate the recovery speed for the disk: T_BaaS=(V_HDD (GB))/(V (GB/hour))
TBaas – is the recovery time (in hours); VHDD – is the disk volume (in GB); V – is the recovery speed (in GB/hour).
To calculate the recovery time for a 500 GB disk:
T_BaaS=(500 (GB))/(200 (GB/h))=2.5(h)
From the data received it can be seen that a 500 GB disk volume can be restored from backup in roughly 2 hours and 30 minutes.