Veritas Technologies LLC v. Veeam Software Corporation

Citations: 835 F.3d 1406; 120 U.S.P.Q. 2d (BNA) 1046; 2016 U.S. App. LEXIS 15978; 2016 WL 4525278Docket: 15-1894

Court: Court of Appeals for the Federal Circuit; August 30, 2016; Federal Appellate Court

Original Court Document: View Document

SummaryOriginalCytator Case TreatmentBackward CitatorAnalysis

EnglishEspañolSimplified EnglishEspañol Fácil

The United States Court of Appeals for the Federal Circuit reviewed the case involving Veritas Technologies LLC and Veeam Software Corporation concerning U.S. Patent No. 7,024,527, which pertains to data restoration methods. Veeam filed a petition for inter partes review, arguing that specific claims of the patent were unpatentable due to prior art. The Patent Trial and Appeal Board (PTAB) instituted the review in April 2014. Veritas, the patent owner, sought to amend the patent by adding claims 26 and 27 if the original claims were found unpatentable. 

In its final April 2015 decision, the PTAB interpreted the claims more broadly than Veritas had argued, determining they covered both file-level and block-level restoration processes. This led to the rejection of the challenged claims based on obviousness under 35 U.S.C. § 103. The Board also denied Veritas's motion to amend, concluding that Veritas failed to demonstrate the patentability of the proposed new claims, as required by the Board.

The appellate court affirmed the PTAB's claim construction as the broadest reasonable interpretation and upheld the obviousness ruling. However, it vacated the denial of Veritas’s motion to amend, finding the Board's reasoning arbitrary and capricious. The case was remanded for the Board to evaluate the patentability of the proposed claims, which Veritas argued had a narrower scope focusing on file-level background restoration.

A data storage device, such as a hard disk, contains blocks that store data for files, with a file system managing the allocation of these blocks. Programs can access data either through file-level access, which involves requesting a file and retrieving its associated blocks via the file system, or through block-level access, where a program directly requests data from a known block. The ’527 patent, titled "Data Restore Mechanism," highlights limitations in current data restoration methods, particularly the lengthy waits applications face for complete file restorations, which can be especially problematic for large databases. The invention aims to create a restore mechanism that minimizes disruptions to active applications and allows for near-instantaneous access to data during the restoration process. It describes a system where a restore application can initiate the restoration of files from backup to primary storage while enabling active applications to request and access specific blocks of data as they are restored. A map links restored data locations to their sources, facilitating immediate restoration of needed blocks if they haven't been restored yet. Claim 20 exemplifies this process, outlining a computer-accessible medium with instructions for a restore application that can concurrently restore and make data accessible to applications during the restoration.

Veeam's corrected petition for inter partes review sought the cancellation of claims 1, 6, 8, 20, and 24 of the ’527 patent, asserting they were anticipated and obvious based on prior art, particularly U.S. Patent Application Publication No. US 2002/0083366 to Ohran, along with a Microsoft Windows NT user guide and U.S. Patent No. 7,234,077 to Curran et al. Ohran's system allows on-demand access to lost data blocks during restoration via two channels: immediate restoration of requested blocks and taking a snapshot of lost data for subsequent restoration. The Patent and Trademark Office Board instituted a review, finding a reasonable likelihood of unpatentability based on the combination of Ohran and other references. 

In response, Veritas argued for a narrow interpretation of the claims, asserting they pertain specifically to file-level restoration, and filed a conditional motion to amend claims to clarify this interpretation. The Board ultimately concluded that the claims are not limited to file-level restoration and can encompass block-level restoration. It found that the claims do not necessitate file-level knowledge, as the identification of 'files' is not required by the claim language or its specification. The Board reasoned that restoring blocks inherently leads to file restoration, thus affirming the obviousness of operating Ohran’s system to restore files.

Veritas’s motion to amend was denied due to insufficient evidence that the newly proposed features in claims 26 and 27 were separately known in the art. The Board indicated that Veritas failed to demonstrate how these features, in combination with existing features, were not present in prior art. Veritas is now appealing the Board's claim construction, its obviousness determination, and the denial of its motion to amend, under the jurisdiction granted by 28 U.S.C. 1295(a)(4)(A).

The Board's claim construction and obviousness determination are upheld based on the broadest-reasonable-interpretation standard, as supported by the precedent set in Cuozzo Speed Techs. LLC v. Lee. There is no expiration date issue for the patent that would necessitate a different claim-construction standard. The Board's construction is reviewed de novo, as there are no disputes regarding the findings or evidence related to extra-patent usage. The ultimate determination of obviousness is also reviewed de novo, while underlying factual findings are assessed for substantial evidence.

The ’527 patent suggests that the restore application operates at the file level, but this indication is insufficient under the broadest-reasonable-interpretation standard to exclude block-level restoration. The patent does not explicitly restrict the restoration process to file-level operations nor does it present aspects of the process that would require such a limitation. Thus, a skilled artisan could reasonably interpret the claims to encompass both file-level and block-level restorations.

Particular claim language, such as "starting a restore of a set of files," implies a file-level operation but does not mandate it. The phrase can be interpreted to allow for block-level restoration, provided it achieves the result of restoring a set of files. Similarly, the requirement regarding "blocks of data of a file" can also be reasonably read to support block-level operations while still achieving file restoration. 

The specification indicates that the restore application generally operates at the file level but does not impose an unreasonable limitation against block-level restoration. The language in the specification discusses restoring files rather than blocks, but does not exclude the possibility of block-level restoration leading to the restoration of complete files. Thus, both the claims and the specification allow for a broader interpretation that includes block-level restoration.

Some embodiments of the invention involve preliminary steps such as retrieving file properties, allocating space for restored files, and mapping data locations on backup and primary storage, which are challenging in block-level restoration contexts. However, these embodiments are specific and do not establish general requirements. The discussion does not imply that the background restore operates differently at file-level versus block-level restoration despite mentioning blocks being retrieved during the process. The specification clarifies that file-level access requires knowledge of the file system, while block-level access does not. It notes that prior systems often required significant wait times for restoration completion. However, it fails to delineate a significant difference between out-of-order block grabbing in file-level versus block-level operations. Consequently, the Board concluded that the broadest reasonable interpretation includes block-level restoration, affirming its analysis of obviousness. Veeam argued that it was evident that blocks in Ohran's process equated to file restoration, a view supported by expert testimony. The Board found that a person skilled in the art would recognize that Ohran’s restoration process could effectively restore files, leading to the conclusion that operating Ohran’s process for file restoration would be obvious. With file-level restoration not deemed a claim requirement, the Board's conclusion on obviousness stands. Furthermore, the Board's denial of Veritas’s motion to amend by adding substitute claims 26 and 27 is reviewed under the Administrative Procedure Act, and it is determined that the Board erred in its reasoning for the denial.

Veritas submitted substitute claims 26 and 27 to clarify the file-level restoration limitation of its patent. Claim 26, substituting for claim 1, specifies that the file set is a subset of previously backed-up files and that the file server receives requests for file portions during restoration. Claim 27, substituting for claim 20, outlines additional preparatory steps and prioritizes the restoration of blocks needed by active applications. Veeam did not contest that these amendments address the prior claim-construction dispute. The Board denied Veritas's motion to amend, asserting that Veritas failed to separately discuss whether the newly added features were known in prior art, concluding that the features were only discussed in combination with other known features. This reasoning was found to be arbitrary and capricious, as Veritas had sufficiently argued that the prior art did not disclose the new features. Veritas specifically highlighted the absence of mechanisms for restoring subsets of files in key references, Ohran and the Windows NT user guide. The court found that Veritas's analysis was clear and did not see the need for further clarification. Consequently, the court vacated the Board’s denial of the motion to amend and remanded the case for a determination of the patentability of claims 26 and 27, while affirming the Board's determinations regarding claims 1, 6, 8, 20, and 24.