Flexible cloud backup service on CloudByte Online Backup and Recovery Business Context Today there are number of cloud-based backup and restore service players which have emerged. Some of these players exclusively cater to individual consumers (offering mostly PC backups), while many of them offer services to consumers and SMEs. The following are key SLAs offered by most of the service providers: * Safe data The data, once stored, would remain persistent with the service provider. * Secured backup Most providers promise that a backup gets stored with the highest level of encryption (e.g. AES 256 bit) * Customization of the backup The size and interval of the backup can be customized as the customer business grows. * DR support The backup is scheduled at regular intervals and the restore operation is performed at regular intervals on the customer DR site. * Flexibility to increase storage space on the fly The end customer can opt for such a service at an additional charge, anytime throughout the service life. Normally providers allow such extensions for business requirements. * Application backups Some of the providers do allow backup of applications such as Oracle Database, MySQL, Microsoft Exchange, etc. * Versioning of backup files, undelete/undo facility The files deleted by the end customer can be undeleted. Changes can be undone. The backup data store maintains multiple versions of the files. * Subscription based business model Most of the providers offer $$ per GB per month per user, where a user can be a LAN user logged into his/her laptop or desktop. * Access Control Set up of permissions for each user of the end customer is available. Suppose company XYZ subscribes to such a service for 10 users u1, u2, u10. Then the backup service provider can allow the backup administrator of XYZ to set different permissions for each of u1, u2, u10.
* Incremental backups The backup is taken only for contents changed from the previous backup. These are contents at the folder level (i.e. addition or deletion of files/folders) OR at the file level (i.e. addition or deletion of contents). * De-duplication Duplicate data/patterns are eliminated from the backup, so that the size of the backup can be reduced further, yielding space saving benefits. Usually this service is offered as offline de-duplication, so that the backup window does not get compromised. * Online sharing of the files/folders backed up - After the backup space is created; the space can be shared with other users, so that they can commonly use it as a shared workplace, in which content changed by one person can be viewed by others instantly. * Maintaining time bound log of changes to the backup space. - The service will maintain the change history of all save/delete/add/create operations occurring in the backup space. Besides these service offerings, the key important business requirement is the ability to provide such services at price points affordable to end customers. The price points from a survey range from $100/TB/month to $200/TB/month. An average end user price offered is in the range of $144/TB/month. Hence, the fundamental requirement becomes how to build storage for such a cloud backup service, such that one can deliver all the SLAs explained above at a price point in the neighborhood of $144/TB/month? Business Challenges 1. End customer feedback indicates key pain points that are yet to be addressed. What follows is the summary of top concerns expressed by the end customers of the cloud backup services. 2. 3. * The backup service provider places hidden limits on the storage capacity, while calling it unlimited. For a large backup, the storage service availability is compromised. 4. 5. * The backup takes lot of time, which is not explicitly stated in the contract with the service provider. 6. 7. * Feature set is good but the price is not affordable Customers do see security, encryption, compression, DR support, access control etc. as very useful features, but mostly these features get compared with direct attached storage which may also be dedicated to the user. In such a context, the price to feature comparison tends to be inevitable; making the cloud based backup solution unattractive.
8. The storage is not reliable. Instances such as loss of a file or folder, which has not been accessed for a while, may be common. This is because while on one hand the service provider tries to provide committed service, on the other hand, it has to also balance it vs. profitability. Current Solutions, their Limitations & Why CloudByte? Having clarified the business context and the business challenges in details, let s have a look at how service providers are attempting to provide the service in the first place and accordingly, what are limitations that directly or indirectly contribute to the challenges outlined above. Let s also look at what are possible alternatives available and their pros and cons. In a very simplistic form, the backup service provider, its end customers and the MSP supporting the backup service provider, can be represented by following block diagram. End customers/ SMEs, End Users of the cloud based backup service Internet Cloud Based Backup Service Provider Management Portal and Datacenter Internet MSP providing DC hosting services to the Cloud Based Backup Service Provider Because it s not cost effective building a datacenter for each service provider, generally Backup Service Providers use datacenter services from MSPs. These services can be delivered as a hosted service or as a cloud service. In the hosted service instance, the MSP would have a dedicated
datacenter setup created and managed for the Backup Service Provider, whereas in the case of a cloud service, then it would be shared, virtual resources made available to the Backup Service Provider. Let s now look at each concern/challenge listed above, and understand how it can be addressed if the Backup Service Provider opts for a hosted service or for a cloud service, to create its service. * Considering the security aspect, the only thing that is guaranteed by the service level is the highest level of encryption of the data. No one is able to provide any security greater than AES 256 bit encryption. Besides, if the Backup Service Provider creates this service on a cloud service from an MSP, then it s up to the capability of the MSP cloud service as to what level of security would be offered. CloudByte creates differentiation in such a scenario. Fundamentally, the CloudByte architecture is based on advanced storage virtualization technology that creates secured muli-tenant isolation for each lease of shared storage infrastructure. Essentially, using the CloudByte multi-tenant storage platform, it s possible for the MSP to create fully secured containers (from access level through to the raw storage level). A secured container can be created either for the Backup Service Provider or for each customer of the Backup Service Provider or one can do a hybrid of both i.e. base package customers of the Backup Service Provider can be securely isolated at the storage level in one or a few containers while premium customers can each be securely isolated in separate containers. * The second key aspect of such a service is in terms of delivering DR capability. I.e. the customer of the Backup Service Provider uses the service first to do the backup of its data from a primary datacenter and also does continuous restore of the backup to its DR site. In the event of disaster, the customer would expect the data to be fully intact and in sync with the primary datacenter. * The storage I/O workload for backup and restore are very different. While backup is writeintensive, the restore is read-intensive. * In the event the Backup Service Provider creates such a service through an MSP, then the MSP needs to provide the capability to create two distinct storage I/O workload SLAs one readintensive and the other write-intensive. As controllers today are not well equipped to handle such scenarios, MSPs would tend to use different controllers to be able to support such a service. This would increase the Capex and Opex of the storage infrastructure. On the other hand, it may be impossible to offer such service on a cloud, as the underlying IT architecture may not be able to provide predictive performance for such requirements. * CloudByte becomes a clear differentiator in handling such a situation. Since CloudByte provides secured isolation through its multi-tenant architecture, the Backup Service Provider can offer this service (at a premium), by configuring 2 different containers for the same customer. One container is used for backup and the other for restore. The Backup container can be optimized to deliver write
IOPS and simultaneously the restore container can be optimized to deliver maximal read IOPS. Both of these configurations can be done on a single instance of the CloudByte controller. Thus while keeping the storage I/O paths and the configuration of backend disk shelves intact, CloudByte delivers predictive performance to these requirements