From aeedc539ab699a76983242687cafc2e3946a4805 Mon Sep 17 00:00:00 2001
From: ArkadyKanevsky <arkady_kanevsky@dell.com>
Date: Mon, 19 Oct 2015 09:49:12 -0500
Subject: [PATCH] DB Hygiene user story

This a proposal for database hygiene for all openstack projects.

Change-Id: I94c205c7a1bb8c43a32e4dc13ff37f8e6a3b666b
---
 user-stories/draft/db-hygiene.rst | 81 +++++++++++++++++++++++++++++++
 1 file changed, 81 insertions(+)
 create mode 100644 user-stories/draft/db-hygiene.rst

diff --git a/user-stories/draft/db-hygiene.rst b/user-stories/draft/db-hygiene.rst
new file mode 100644
index 0000000..a280719
--- /dev/null
+++ b/user-stories/draft/db-hygiene.rst
@@ -0,0 +1,81 @@
+Database cleanup of deleted object
+============================================
+
+*Problem description*
+---------------------
+Each operator of an OpenStack cloud needs the ability to clean up the OpenStack
+database of objects which have been deleted. Currently a new record is created in
+the OpenStack database when an object (project, user, VM, network, volume, swift
+object, etc.) is created. When an object is deleted its record in the database
+remains but is marked as deleted.  As an OpenStack cloud stays in operation over
+time, the number of records in the database fills with deleted object records and
+soon the database becomes too large. This can result in the database consuming
+significant resources on the controller node(s) and impacting responsiveness of
+the database, even bringing down controller(s).
+
+While a record for deletion of an object is needed for cloud governance it is
+not a requirement for the database to retain this record. Thus, some tool is
+needed for OpenStack operators to cleanup the database of records from deleted
+objects.
+
+* As part of a PoC of an OpenStack cloud, an operator needs a tool to purge the
+database of deleted records. This is needed in order to be able to rerun the same
+test tool that would create objects every run as well as testing actual growth of
+the database for operational conditions not OpenStack implementation artifacts.
+
+* As an operator of an OpenStack cloud I need an ability to remove records of
+deleted objects from the database after storing those records for audit purposes.
+
+* As a cloud deployment engineer, I need to be able to re-run the same tests
+(with same objects) in a repeatable manner so that I can have a high certainty
+in the outcome of my proof of concept and cloud functionality.
+
+* As a cloud deployment engineer, I need only the appropriate records in my
+database so that I can complete my upgrade in the allocated down time.
+
+Usage Scenarios Examples
+------------------------
+TBD
+
+Opportunity/Justification
+-------------------------
+DB hygiene is required for handling OpenStack performance, operational and
+upgrade issues. This ensures that historical records of deleted items are not
+impacting operational performance and such deleted items are not polluted by
+upgrades.
+
+
+Related User Stories
+--------------------
+Nova specs:
+* https://review.openstack.org/#/c/184645/
+* https://review.openstack.org/#/c/184637/
+* https://review.openstack.org/#/c/137669/
+
+Cinder blueprint:
+* https://blueprints.launchpad.net/cinder/+spec/db-cleanup
+
+*Requirements*
+--------------
+* Operator should be able to specify which policy to apply for deleted objects
+* Operator should be able to specify which policy to apply for different tenants
+and sub-tenants.
+* At least two policies must be supported: Policy 1 - Archive the records in
+other persistent storage for a specific interval duration; Policy 2 - Remove
+the records from database permanently.
+
+*Gaps*
+------
+None currently known.
+
+*Affected By*
+-------------
+None.
+
+*External References*
+---------------------
+None.
+
+Glossary
+--------
+None.