Perfect Search's Answer to Backup's Retrieval Problem

The Problem

Today's backup technologies are great for capturing data and moving it to a safe, secondary location. The depth and breadth of backup features is impressive. However, the entire industry suffers an embarrassing Achilles heel: retrieval and recovery is primitive at best. When the CEO calls IT in a panic to report that she accidentally deleted her investor presentation sometime late last week, the IT staff's blood pressure goes through the roof. Assuming their catalog is accurate and up-to-date, and assuming their backup media is reliable, they can mount the most recent snapshot of the file server the CEO was using. But will it yield the document they need? How can they even identify the desired file, if the CEO doesn't remember what she called it? What if the CEO's final changes were only in a version saved on her laptop? How can IT be confident of a response if they have to solve the emergency in just a few minutes?

Ideally, IT would address this problem by asking the backup system for a list of files that match the CEO's description--perhaps .ppt and .pptx files modified in the past week that match key words "strategic" and "investor".

Unfortunately, that ideal is pie-in-the-sky. Today's backup catalogs are not granular enough to find what IT needs, and even if they were, they don't support straightforward google-style queries. This is not because backup engineers lack vision or commitment to address customer requirements; it's because of inherent limits in the indexing technology used by backup engines. Specifically, standard indexing technology:

- Indexes files too slowly to make indexing of document content practical.
- Uses prohibitive amounts of space.
- Bogs down when data sets grow too large.
- Is difficult to adapt to the notion of incremental snapshot layers.
- Is optimized for SQL-style usage, requiring accurate knowledge of document attributes.

The Solution

Perfect Search has a powerful, patented search platform that addresses each of these issues. Perfect Search indexing is several times faster than traditional approaches; content indexing is straightforward and modest in its storage requirements. Perfect Search can index and query an order of magnitude more records on the same hardware. Perfect Search indexes can be layered just like incremental backups--and they can be queried with simple google-style search phrases.

When Perfect Search's technology is combined with state-of-the-art backup features, customers get the best of both worlds: rock-solid preservation of data that matters, and the confidence of pain-free recovery.

How It Works

Perfect Search can be configured to index a data set in real-time, accumulating changes as they occur, and peering into as much or as little of the internal content of documents as desired. Alternatively, Perfect Search can quickly create an index in a single pass when a snapshot is taken. In either case, the Perfect Search index that corresponds to a snapshot is saved to a self-contained index file that matches the data in a full or incremental backup. This index can be stored on the backup media itself, or it can be moved to a central repository for easy search federation--or both strategies can be used.

The incremental layers of Perfect Search indexes can also be merged or deleted. This allows a backup system to maintain correspondence between indexes and backups as routine maintenance alters retained history, removing a major stumbling block to flexible retrieval.

At query time, the user provides search criteria on a web page; the experience is much like a search in any internet search engine. Underneath, the query engine transparently consults each index layer to generate a composite, ranked set of hits. Each hit comes with a document excerpt and other key descriptors that allow the user to browse, before any data is even fetched from the backup. Hits also have a ready-made URL that maps them back to the backup they "live" in; once a hit is selected, the backup engine invokes its own mount-and-retrieve logic to restore.

Perfect Search's technology is sold as an SDK to backup providers. Perfect Search can also implement a custom search solution on top of traditional tape-, disk-, or cloud-based backup systems in a professional services engagement.

Benefits

A deployed backup+Perfect Search solution takes the panic out of recovery. Backup administrators quickly discover that they can find what they need. Perhaps more interestingly, they have the freedom to offer self-serve searches to the users they support. Those users already know how to use a simple search box; the need for intervention by an IT professional is governed only by audit or procedural considerations.

Perfect Search's technology also makes backed up data discoverable in new ways. Regulatory compliance, auditing, and e-discovery use cases can be addressed by generating reports about hits on a particular topic, date range, document type, and so forth.