In some cases, both kernel read-ahead and gluster read-ahead mechanisms cause a large increase in the amount of data read during sequential read operations. In these cases, performance for sequential reads, especially on dispersed volumes, is significantly worse. If you are affected by this issue, you can work around the problem by manually disabling gluster read-ahead and io-cache behavior using the following commands. # gluster volume set <volname> read-ahead off # gluster volume set <volname> io-cache off