asked on

MYSQL Group together query

Hi Everyone,

I am trying to put together a query that does the following:

I have a table with 512 entries in it (initially, this will increase with records at we add to it)

This table has columns id, ip_address, in_use where (id is auto_incrementing and primary) and in_use is bool true / false.

Basically I have a list of addresses in this (like I said initially it's got 512)

I need to ask mysql to return rows not in use, but group them together by a set number (ie I need 11 free addresses),

However, this system will have rows marked false on the in_use col right next to others with true.

I need a way to grab a set amount of IP's the id being broken sequentially... example

ids => in_use

1 => false

2 => false

3 => false

4 => false

5 => false

6 => false

7 => false

8 => false

9 => true

10 => true

11 => false

12 => false

13 => false

14 => false

15 => false

16 => false

17 => false

18 => false

19 => false

20 => false

21 => false

22 => false

23 => false

Now if I query for 11 free addresses, closest to the top, it will bring back the list of id's 1 to 12 skipping over 9, this won't work for what I am trying to do :(

What I would like it to do is realize that 9 is in use so cannot use that, thus cannot make a sequential group and pick from 11 to 22... as that is the next available sequential set.

Is there any way to structure a query for this? Is it even possible in MySQL?

Your help is kindly appreciated, Karl.

Aaron Tomosky

Without a fully flushed out solution I think I can say it may be possible and get you started.
If you add a rowcount column to the query output and then take the last part of the ip and subtract the rownumber, that should give you a constant number for contiguous ip addresses. Group by that column and add a count to see contiguous blocks. Change the ip to max(ip), ad a min(ip) and you should get start and end and counts of all the blocks. I won't be back at a PC until tomorrow to test all this but I think I will work and someone may be able to write it up before then ifyou can't figure it out.

crucial_paradigm

ASKER

Thanks Aaron!

I hope to see some type of demonstration of this soon, I am a bit baffled by your terminology so I will await your thoughts

I am doing this currently in the application level, but practically dropping a few thousand rows into array returns is not a good idea....

I look forward to hearing from you!
--Karl.

SOLUTION

Aaron Tomosky

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

ASKER CERTIFIED SOLUTION

Sharath S

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

kristinalim

If you'll be getting more queries than updates, you might want to consider a caching mechanism:

1. Add column for tracking number of not in use IP addresses from current record until the next IP address not in use. For example:

id  in_use  contiguous_unused
1   true      1
2   false     0
3   true      1
4   true      2
5   false     0

Open in new window

2. Then, when marking a record with ID x not in use:

# Get first used IP address after current record with ID x.
next_used_id : SELECT MIN(id) FROM ip_addresses WHERE in_use IS TRUE AND id > x;
# Get contiguous_unused for current record.
current_contiguous_unused : SELECT COUNT(*) + 1 FROM ip_addresses WHERE in_use IS FALSE and id > x and id < next_used_id;
# Set contiguous_unused of current record.
UPDATE ip_addresses SET contiguous_unused = current_contiguous_unused WHERE id = x;
# Get last used IP address before current record with ID x.
last_used_id : SELECT MAX(id) FROM ip_addresses WHERE in_use IS TRUE AND id < x;
# Increment contiguous_unused by current_contiguous_unused for all records after last_used_id until previous record.
UPDATE ip_addresses SET contiguous_unused = contiguous_unused + current_contiguous_unused WHERE id > last_used_id AND id < x;

Open in new window

Essentially, you would be adjusting the cached count for records with affected counts.

3. When marking a record with ID x as in use:

# Get last used IP address before current record with ID x.
last_used_id : SELECT MAX(id) FROM ip_addresses WHERE in_use IS TRUE AND id < x;
# Set contiguous_unused of current record.
UPDATE ip_addresses SET contiguous_unused = 0 WHERE id = x;
# Decrement contiguous_unused by 1 for all records after last_used_id until previous record.
UPDATE ip_addresses SET contiguous_unused = contiguous_unused - 1 WHERE id > last_used_id AND id < x;

Open in new window

4. A new record will just have 0 for contiguous_unused if in_use and 1 otherwise.
5. Selecting first set of 11 contiguous records:

SELECT MIN(id) FROM ip_addresses WHERE contiguous_unused > 11;

Open in new window

This is a rough algorithm. Of course you will have to catch cases when last used or next used record is not existent (start or end of table) and optimize for mass inserts/updates.

Just a possibility.

kristinalim

Correction

5. Selecting first set of 11 contiguous records:

SELECT MIN(id) FROM ip_addresses WHERE contiguous_unused >= 11;

Open in new window

You can have flexible queries with this too, if you want to get the smallest unused range or even middle depending on your method for space optimization.

crucial_paradigm

ASKER

Thank you all for your replies,

Because of this I have now been able to form my database table and use it correctly!

--Karl