[ https://issues.apache.org/jira/browse/CASSANDRA-3620?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Dominic Williams updated CASSANDRA-3620: ---------------------------------------- Description: Here is a proposal for an improved system for handling distributed deletes. h2. The Problem Repair has issues: * Repair is expensive anyway * Repair jobs are often made more expensive than they should be by other issues (nodes dropping requests, hinted handoff not working, downtime etc) * Repair can often itself fail and need restarting, especially in cloud environments where a network issue might make a node disappear from the ring for a brief moment * When you fail to run repair within GCSeconds, either because you are dumb or because of issues with Cassandra, deleted data can reappear * If you cannot run repair and have to increase GCSeconds, tombstones can start overloading your system Because of the foregoing, in high throughput environments you often cannot make repair a cron job. You prefer to keep a terminal open and run repair jobs one by one, making sure they succeed and keeping and eye on overall load so you don't impact your system. This isn't great, and it is made worse if you have lots of column families or have to run a low GCSeconds on a column family to reduce tombstone load. You know that if you don't manage to run repair with the GCSeconds window, you are going to hit problems, and this is the Sword of Damocles over your head. Running repair to deal with missing writes isn't so important, since QUORUM reads will always receive data successfully written with QUORUM. Ideally there should be no ongoing requirement to run repair to avoid data loss, and no GCSeconds. Repair should be an optional maintenance utility used in special cases, or to ensure ONE reads get consistent data. h2. "Reaper Model" Proposal # Tombstones do not expire, and there is no GCSeconds # Tombstones have associated ACK lists, which record the replicas that have acknowledged them # Tombstones are only deleted (or marked for compaction) when they have been acknowledged by all replicas # When a tombstone is deleted, it is added to a fast "relic" index of MD5 hashes of cf-key-name[-subName]-ackList. The relic index makes it possible for a reaper to acknowledge a tombstone after it is deleted # Background "reaper" threads constantly stream ACK requests to other nodes, and stream back ACK responses back to requests they have received (throttling their usage of CPU and bandwidth so as not to affect performance) # If a reaper receives a request to ACK a tombstone that does not exist, it creates the tombstone and adds an ACK for the requestor, and replies with an ACK NOTES * The existence of entries in the relic index do not affect normal query performance * If a node goes down, and comes up after a configurable relic entry timeout, the worst that can happen is that a tombstone that hasn't received all its acknowledgements is re-created across the replicas when the reaper requests their acknowledgements (which is no big deal since this does not corrupt data) * Since early removal of entries in the relic index does not cause data loss, it can be kept small, or even kept in memory * Simple to implement and predictable h3. Benefits * Operations are finely grained (reaper interruption is not an issue) * The labour & administration overhead associated with running repair can be removed * Reapers can utilize "spare" cycles and run constantly in background to prevent the load spikes and performance issues associated with repair * There will no longer be the threat of data loss if repair can't be run for some reason (for example because of a new adopter's lack of Cassandra expertise, a cron script failing, or Cassandra bugs preventing repair being run etc) * Deleting tombstones earlier, thereby reducing the number involved in query processing, will often dramatically improve performance was: Here is a proposal for an improved system for handling distributed deletes. h2. The Problem Repair has issues: * Repair is expensive anyway * Repair jobs are often made more expensive than they should be by other issues (nodes dropping requests, hinted handoff not working, downtime etc) * Repair can often itself fail and need restarting, especially in cloud environments where a network issue might make a node disappear from the ring for a brief moment * When you fail to run repair within GCSeconds, either because you are dumb or because of issues with Cassandra, deleted data can reappear * If you cannot run repair and have to increase GCSeconds, tombstones can start overloading your system Because of the foregoing, in high throughput environments you often cannot make repair a cron job. You prefer to keep a terminal open and run repair jobs one by one, making sure they succeed and keeping and eye on overall load so you don't impact your system. This isn't great, and it is made worse if you have lots of column families or have to run a low GCSeconds on a column family to reduce tombstone load. You know that if you don't manage to run repair with the GCSeconds window, you are going to hit problems, and this is the Sword of Damocles over your head. Running repair to deal with missing writes isn't so important, since QUORUM reads will always receive data successfully written with QUORUM. Ideally there should be no ongoing requirement to run repair to avoid data loss, and no GCSeconds. Repair should be an optional maintenance utility used in special cases, or to ensure ONE reads get consistent data. h2. "Reaper Model" Proposal # Tombstones do not expire, and there is no GCSeconds # Tombstones have associated ACK lists, which record the replicas that have acknowledged them # Tombstones are only deleted (or marked for compaction) when they have been acknowledged by all replicas # When a tombstone is deleted, it is added to a fast "relic" index of MD5 hashes of cf-key-name[-subName]-ackList. The relic index makes it possible for a reaper to acknowledge a tombstone after it is deleted # Background "reaper" threads constantly stream ACK requests to other nodes, and stream back ACK responses back to requests they have received (throttling their usage of CPU and bandwidth so as not to affect performance) # The relic index is scavenged according to some configurable period # If a reaper receives a request to ACK a tombstone that does not exist, it creates the tombstone and adds an ACK for the requestor, and replies with an ACK NOTES * The existence of entries in the relic index do not affect normal query performance * If a node goes down, and comes up after the configurable relic entry timeout, the worst that can happen is that a tombstone that hasn't received all its acknowledgements is re-created across the replicas (which is no big deal since does not corrupt data) h3. Benefits * The labour/administration overhead associated with running repair will be removed * The reapers can utilize "spare" cycles and run constantly to prevent the load spikes and performance issues associated with repair * There will no longer be the risk of data loss if repair can't be run for some reason (for example because of a new adopter's lack of Cassandra expertise, a cron script failing, or Cassandra bugs preventing repair being run etc) * Reducing the average number of tombstones databases carry will improve performance, sometimes dramatically > Proposal for distributed deletes - use "Reaper Model" rather than GCSeconds > and scheduled repairs > ------------------------------------------------------------------------------------------------- > > Key: CASSANDRA-3620 > URL: https://issues.apache.org/jira/browse/CASSANDRA-3620 > Project: Cassandra > Issue Type: Improvement > Components: Core > Affects Versions: 1.0.5 > Reporter: Dominic Williams > Labels: GCSeconds,, deletes,, distributed_deletes,, > merkle_trees, repair, > Fix For: 1.1 > > Original Estimate: 504h > Remaining Estimate: 504h > > Here is a proposal for an improved system for handling distributed deletes. > h2. The Problem > Repair has issues: > * Repair is expensive anyway > * Repair jobs are often made more expensive than they should be by other > issues (nodes dropping requests, hinted handoff not working, downtime etc) > * Repair can often itself fail and need restarting, especially in cloud > environments where a network issue might make a node disappear > from the ring for a brief moment > * When you fail to run repair within GCSeconds, either because you are dumb > or because of issues with Cassandra, deleted data can reappear > * If you cannot run repair and have to increase GCSeconds, tombstones can > start overloading your system > Because of the foregoing, in high throughput environments you often cannot > make repair a cron job. You prefer to keep a terminal open and run repair > jobs one by one, making sure they succeed and keeping and eye on overall load > so you don't impact your system. This isn't great, and it is made worse if > you have lots of column families or have to run a low GCSeconds on a column > family to reduce tombstone load. You know that if you don't manage to run > repair with the GCSeconds window, you are going to hit problems, and this is > the Sword of Damocles over your head. > Running repair to deal with missing writes isn't so important, since QUORUM > reads will always receive data successfully written with QUORUM. > Ideally there should be no ongoing requirement to run repair to avoid data > loss, and no GCSeconds. Repair should be an optional maintenance utility used > in special cases, or to ensure ONE reads get consistent data. > h2. "Reaper Model" Proposal > # Tombstones do not expire, and there is no GCSeconds > # Tombstones have associated ACK lists, which record the replicas that have > acknowledged them > # Tombstones are only deleted (or marked for compaction) when they have been > acknowledged by all replicas > # When a tombstone is deleted, it is added to a fast "relic" index of MD5 > hashes of cf-key-name[-subName]-ackList. The relic index makes it possible > for a reaper to acknowledge a tombstone after it is deleted > # Background "reaper" threads constantly stream ACK requests to other nodes, > and stream back ACK responses back to requests they have received (throttling > their usage of CPU and bandwidth so as not to affect performance) > # If a reaper receives a request to ACK a tombstone that does not exist, it > creates the tombstone and adds an ACK for the requestor, and replies with an > ACK > NOTES > * The existence of entries in the relic index do not affect normal query > performance > * If a node goes down, and comes up after a configurable relic entry timeout, > the worst that can happen is that a tombstone that hasn't received all its > acknowledgements is re-created across the replicas when the reaper requests > their acknowledgements (which is no big deal since this does not corrupt data) > * Since early removal of entries in the relic index does not cause data loss, > it can be kept small, or even kept in memory > * Simple to implement and predictable > h3. Benefits > * Operations are finely grained (reaper interruption is not an issue) > * The labour & administration overhead associated with running repair can be > removed > * Reapers can utilize "spare" cycles and run constantly in background to > prevent the load spikes and performance issues associated with repair > * There will no longer be the threat of data loss if repair can't be run for > some reason (for example because of a new adopter's lack of Cassandra > expertise, a cron script failing, or Cassandra bugs preventing repair being > run etc) > * Deleting tombstones earlier, thereby reducing the number involved in query > processing, will often dramatically improve performance -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira