Paper Title: SVL: Storage Virtualization Engine Leveraging DBMS Technology

Author(s): Lin Qiao, Balakrishna R. Iyer, Divyakant Agrawal, Amr El Abbadi

1) Is the paper technically correct?
 [X] Yes
 [ ] Mostly (minor flaws, but mostly solid)
 [ ] No

2) Originality
 [X] Very good (very novel, trailblazing work)
 [ ] Good 
 [ ] Marginal (very incremental)
 [ ] Poor (little or nothing that is new)

3) Technical Depth
 [ ] Very good (comparable to best conference papers)
 [X] Good (comparable to typical conference papers)
 [ ] Marginal depth
 [ ] Little or no depth

4) Impact/Significance
 [ ] Very significant
 [X] Significant
 [ ] Marginal significance.
 [ ] Little or no significance.

5) Presentation
 [ ] Very well written
 [X] Generally well written
 [ ] Readable
 [ ] Needs considerable work
 [ ] Unacceptably bad

6) Overall Rating
 [ ] Strong accept (very high quality)
 [X] Accept (high quality - would argue for acceptance)
 [ ] Weak Accept (marginal, willing to accept but wouldn't argue for it)
 [ ] Weak Reject (marginal, probably reject)
 [ ] Reject (would argue for rejection)

7) Summary of the paper's main contribution and rationale
   for your recommendation. (1-2 paragraphs)

This paper contributes a novel approach to storage management by using a relational
database management system for block virtualization. The motive behind this approach
is to take advantage of the mature functionality of DBMS for use as an expressive,
fault-tolerant, and secure virtualized I/O layer between applications and physical
disks. The proposed method uses a generic, "off-the-shelf", DBMS as the management
tool for virtualizing physical blocks; however, the actual implemented system required
extensive modifications to the DBMS kernel in order to overcome performance issues.
With the modified system, the performance is comparable to other existing technologies
which serve a similar purpose (VSD), and additional DBMS functionality is gained (such
as encryption for cells, columns, or tables, table based compression, etc...). The
paper presents detailed descriptions of how to design such a system, including the
most significant task of mapping I/O requests (ex SCSI) to SQL queries for the DBMS
to process. The paper also validates the idea with empirical results showing the
comparable performance to VSD. Only the modified DBMS version has acceptable performance.
It is because of the novelty of the idea and thoroughness of description and evaluation
that I recommend this paper for acceptance.



8) List 1-3 strengths of the paper.  (1-2 sentences each,
identified as S1, S2, S3.)

S1. Novel and interesting use for traditional relational databases.


9) List 1-3 weaknesses of the paper (1-2 sentences each,
identified as W1, W2, W3.)

W1. Actual system needs extensive modifications to the DBMS (which goes against the advertised "how to take a general purpose commercial RDBMS, rather than a specialized solution, to support block storage management"). The Vanilla system has poor performance which suggests that an off-the-shelf DBMS could not be used in practice.


10) Detailed comments for authors.

This paper presents an interesting use for relational databases and provides an interesting
solution to the problem of physical block virtualization. It would interesting to see
additional benefits of the DBMS functionality, such as rollbacks from the logged transactions
to restore old/lost/corrupted files. Also, the mentioned future work of operating in a fully
distributed environment seems like a promising application of this method.