Every organization needs to be sure that its data is backed up, without depending on end-users to do it. The easiest way to secure those endpoints and their data — quickly, securely, and efficiently — is in the cloud. And it’s particularly important when you need to back up thousands of devices at the same time.
Backing up data from enterprise endpoints — corporate laptop computers, mobile devices, or anything on which the organization’s data can reside — can be a nightmare of uncertainty and risk. For example, users often backup their laptops (if they back them up at all) on flash drives, consumer file sharing services, or even by emailing attachments to themselves. But flash drives are too hard to identify and track. Users lose them or have a drawerful of flash drives with no efficient way to identify them. File sharing services for consumers are handy but poorly secured, and emailing simply adds to the approximately 80% worth of data that are duplicated.
In contrast, backing up endpoints to a secure and stable cloud solves these problems. A stable cloud offers up to 99.5% availability and 99.99999% durability (eleven 9’s) of guaranteed service level agreements (SLAs) depending on your service level agreements and contract pricing. Extreme scalability, security, reduced infrastructure costs, and management time round out the business case.
However, sometimes IT needs to backup, restore, upgrade, or migrate many endpoints at once. Depending on the WAN to transport data to and from thousands of endpoints can create unacceptable demands on the backup infrastructure. Long transport times and high bandwidth demands can overrun backup windows, slow down applications, and cause SLAs to fail.
Combining Cloud and Cache
Here’s the best solution to this big problem: combine the cloud data repository and local cache — add dedupe and WAN optimization — and you have the high performance you need for large-scale backup and recovery. There are five critical requirements for this level of performance: speedy backup, fast restores on demand, controlling expenses at the caching storage layer, global dedupe, and optimizing the WAN.
- High performance cloud backup. IT commonly schedules backup for off-peak hours, but even local LAN backups overrun backup time windows. Using the WAN to transport data to the cloud multiplies the problem. IT should combine off-peak backup with global dedupe, bandwidth throttling, and WAN optimization to efficiently backup data. At the highest levels of efficiency, this enables IT to run large baseline backups and even data migrations on tens of thousands of user devices without paralyzing performance.
- Fast restores on demand. Restores cannot wait for off-peak hours; they have to be ready whenever a user need occurs. Optimizing WAN transfers helps considerably. Multi-threading for parallel restores operations allows IT to efficiently restore even large volumes within recovery time objectives (RTO).
- Optimize local cache storage for capacity. The amount of storage that the cache server requires can add ongoing expense to a cloud backup project. Optimal cloud cache solutions manage storage capacity needs and do not require an ongoing storage investment.
- Optimize deduplication for cloud backup. Dedupe is a critical component in any backup scenario but is an intensive operation with high overhead. Large amounts of data lengthen dedupe times, which is a noticeable factor. But the real issue is that up to 80% of enterprise data are copies of other documents. (How many Grumpy Cat photos are on your users’ systems?!) Global deduplication vastly improves dedupe efficiency by identifying and removing copies across the global data store, not simply on each device.
- WAN optimization. Optimizing the WAN also speeds up data transfers by bandwidth throttling, multi-threading, and seamlessly resuming backup following a network interruption.
Naturally, at Druva we feel our product design fulfills all these goals admirably.
Druva CloudCache is an optional, innovative software appliance for high performance cloud backup for endpoint devices. Druva inSync customers deploy the CloudCache appliance on-site, which changes the backup process to accommodate greater LAN speeds.
When an inSync client is working remotely, Druva inSync runs global dedupe from the Druva Cloud and backs up the data as always. But when the device networks into the LAN, or when IT initiates data transfers on-site to the endpoints, the process changes.
The Druva Cloud still receives the backup request and stores the metadata, which it manages for consistent snapshots and global dedupe. But instead of initiating backup directly over the WAN, the cloud routes the request to the CloudCache server. The backup data travels quickly over the LAN to the CloudCache, which performs global dedupe using the Cloud’s dedupe engine. The server efficiently syncs only the deduped blocks to the cloud, immediately or on IT’s schedule. WAN transfer speeds are very fast.
The restore operation is just as efficient. When an inSync Cloud client device is on the LAN and initiates a restore, the request travels directly to Druva’s Cloud. The Cloud immediately calculates the correct deduped blocks and returns the restored data to CloudCache. The caching server quickly restores the data to the device over the LAN. If applications and users need even faster service, IT can keep the latest snapshotson the local cache for immediate restores.
The operation also works for large-scale device refreshes or operating system migration. Using CloudCache, IT can preload the data into the cache for select user groups and schedule the refresh or migration for off-peak hours.
CloudCache is the answer to several big IT challenges: backing up a large volume of initial data, other large data transfer events like device refreshes or OS migrations, accelerating ongoing backup and restore, and protecting WAN connectivity for backup operations and other applications.
Find out more about CloudCache and how it can turbocharge your environment in CloudCache: On-Site Cache Appliance for Turbocharged Backups & Restores.