Hello reader,
prerequisite of this blog :
So here goes the story ,
Intro
We know redis is a simple key value in-memory database. However, there was an instance where I wanted to have a tree but still get the performance of redis.
I thought of sharing the solution if anyone would like to use it for their reference.
Idea's intro (Organizing your data )
In a way, the solution is pretty simple. Below is the image that would help you understand the solution.
Every node has a name without any special characters, and it should follow a specific format.
every entry in redis as you might know is key , value
and that we represent as nodename , nodevalue
Every node has value (say json as string)
key = node name
value = json string (that is the value we want to save in a node )
node name organized as below :
#parent1#parent2#parent3# ....#leafnode#
How to traverse
search string if used as below you can parse as a tree (used regex as supported by redis):
A#* all children of A
A#B#*
all children of B which is a child of node A and A is root
if the search would have been
*#B#*
all Children of node B and (B this time need not be root)
*#B#
Node B which might have parent
This is a pretty simple hack and works on an idea of creating node names with combination of node and edge names.
Statutory Warning
Few things to think before actually using this solution:
If you have the independence, consider using solutions as graphical DBs such as like Neo4J orientdb, these are pretty good ones and we have a very vast libraries built around themselves. Though I used above solution for a brief period , I could very easily convince that graphical DB is much better solution.
It's a tree and not graph, so the chances of it's usage are limited.
And above all , it's just a hack :) beware of hacks always !!
prerequisite of this blog :
- Basics of graphs and trees
- minimal exposure to Redis
So here goes the story ,
Intro
We know redis is a simple key value in-memory database. However, there was an instance where I wanted to have a tree but still get the performance of redis.
I thought of sharing the solution if anyone would like to use it for their reference.
Idea's intro (Organizing your data )
In a way, the solution is pretty simple. Below is the image that would help you understand the solution.
Every node has a name without any special characters, and it should follow a specific format.
every entry in redis as you might know is key , value
and that we represent as nodename , nodevalue
Every node has value (say json as string)
key = node name
value = json string (that is the value we want to save in a node )
node name organized as below :
#parent1#parent2#parent3# ....#leafnode#
How to traverse
search string if used as below you can parse as a tree (used regex as supported by redis):
A#* all children of A
A#B#*
all children of B which is a child of node A and A is root
if the search would have been
*#B#*
all Children of node B and (B this time need not be root)
*#B#
Node B which might have parent
This is a pretty simple hack and works on an idea of creating node names with combination of node and edge names.
Statutory Warning
Few things to think before actually using this solution:
If you have the independence, consider using solutions as graphical DBs such as like Neo4J orientdb, these are pretty good ones and we have a very vast libraries built around themselves. Though I used above solution for a brief period , I could very easily convince that graphical DB is much better solution.
It's a tree and not graph, so the chances of it's usage are limited.
And above all , it's just a hack :) beware of hacks always !!
Please comment and correct if you feel something feels not correct.
