- ...Disk
- This work was supported in part by the Defense Advanced
Research Projects Agency (DABT63-96-C-0056), the National Science
Foundation (CDA 9401156), California MICRO, the AT&T Foundation,
Digital Equipment Corporation, Hewlett Packard, IBM, Intel, Sun
Microsystems, and Xerox Corporation. Anderson was also supported by a
National Science Foundation Presidential Faculty Fellowship.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
- ...Wang
- Computer Science Division,
University of California, Berkeley,
{rywang,pattrsn}@cs.berkeley.edu
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
- ...Anderson
- Department of Computer Science and
Engineering, University of Washington, Seattle,
tom@cs.washington.edu
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
- ...(80%)
- The VLD
latency in this case is measured immediately after running a compactor,
as explained in Section 5.5.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
- ...time
- In this case, because the NVRAM size is
larger than the available free space, the cleaner still needs to run
during flushing to reclaim the free space created during flushing.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
