hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Henning Blohm <henning.bl...@zfabrik.de>
Subject Re: Using HBase timestamps as natural versioning
Date Fri, 30 Aug 2013 12:42:47 GMT
Was gone for a few days. Sorry for not getting back to this until now. 
And thanks for adding to the discussion!

The time used in the timestamp is the "natural" time (in ms resolution) 
as far as known. I.e. in the end it is of course some machine time, but 
the trigger to choose it is some human interaction typically. So there 
is some natural time to events that update a row's data.
If timestamps happen to differ just by 1 ms, as unlikely as that may be, 
this would still be valid.
And the timestamp is always set by the client (i.e. the app server) when 
performing an HBase put. So it's never the region server time or 
something slightly arbitrary.

To recap: The data model (even before mapping to HBase) is essentially

ID -> ( attribute -> ( time -> value ))

(where ID is a composite key consisting of some natural elements and 
some surrogate part).

An event is something like "at time t, attribute x of  ID attained value 
z".

Events may enter the system out of timely order!

Typical access patterns are:

(R1) "Get me all attributes of ID at time t"
(R2) "Get me a trails of attribute changes between time t0 and t1"
(W1) "Set x=z on ID for time t"

As said, currently we store data almost exactly the way I described the 
model above (and probably that's why I wrote it down the way I did) 
using the HBase time stamp to store to time dimension.


Alternative: Adding the time dimension to the row key
-----------

That would mean: ID/time -> (attribute -> value)

That would imply to either have copies of all (later) attribute values 
in all (later) rows or to only put deltas and to scan over rows to 
collect attribute values.

Let's assume the latter (for better storage and writing performance).

Wouldn't that mean to rebuild what HBase does? Is there nothing HBase 
does more efficient when performing R1 for example?

I.e: Assume I want to get the latest state of row ID. In that case I 
would need to scan from ID/0 to ID/<now> (or reverse) to fish for all 
attribute values (assuming I don't know all expected attributes 
beforehand). Is that as efficient as an HBase get with max versions 1 
and <now> as time stamp?

Thanks,
Henning


On 08/21/2013 01:11 PM, Michael Segel wrote:
> I would have to disagree with Lars on this one...
>
> Its really a bad design.
>
> To your point, your data is temporal in nature. That is to say, time is an element of
your data and it should be part of your schema.
>
> You have to remember that time is relative.
>
> When a row is entered in to HBase, which time is used in the timestamp?
> The client(s)? The RS?  Unless I am mistaken or the API has changed, you can set up any
arbitrary long value to be the timestamp for a given row/cell.
> Like I said, its relative.
>
> Since your data is temporal what is the difference if the event happened at TS xxxxxxxx10
xxxxxxxxx11 (the point is that the TS is different by 1 in the least significant bit)
> You could be trying to reference the same event.
>
> To Lars point, if you make time part of your key, you could end up with hot spots. It
depends on your key design. If its the least significant portion of the key, its less of an
issue. (clientX | action | TS) would be an example that would sort the data by client, by
action type, then by time stamp.  (EPOCH - TS ) would put the most current first.
>
> When you try to take a short cut, it usually will bite you in the ass.
>
> TANSTAAFL applies!
>
> HTH
>
> -Mike
>
> On Aug 11, 2013, at 12:21 AM, lars hofhansl <larsh@apache.org> wrote:
>
>> If you want deletes to work correctly you should enable KEEP_DELETED_CELLS for your
column families (I still think that should be the default anyway).
>> Otherwise time-range queries will not be correct w.r.t. deleted data (specifically
you cannot get back at deleted data even if you specify a time range before the delete and
even if you column family as unlimited versions).
>>
>>
>> Depending on what your typical queries are, you might run into performance issues.
HBase sorts all versions of a KeyValue adjacent to each other.
>> If you now want to query only along the latest data (the last version), HBase will
have to skip a lot of other versions. In the worst case the latest version of all KeyVales
are on separate (HFile) blocks.
>>
>> The question of whether to use the builtin timestamps or model the time as part of
the row keys (or even a time-column), is an interesting one.
>> Generally the row-key identifies your row. If you want a new row for each TS in your
logical model you should manage the time dimension yourself.
>> Otherwise if you identities (i.e. row) with many versions, the builtin TS might be
better.
>>
>> -- Lars
>>
>> ________________________________
>> From: Henning Blohm <henning.blohm@zfabrik.de>
>> To: user <user@hbase.apache.org>
>> Sent: Saturday, August 10, 2013 6:26 AM
>> Subject: Using HBase timestamps as natural versioning
>>
>>
>> Hi,
>>
>> we are managing some naturally time versioned data in HBase. That is,
>> there are change events that have a specific time set and when such
>> event is handled, data in HBase, pertaining to the exact same point in
>> time, is updated.
>>
>> So far we are using HBase time stamps to model the time dimension. All
>> columns have unlimited number of versions. That worked ok so far, and
>> HBase's way of providing access to data at a given time or time range
>> seemed a natural fit.
>>
>> We are aware of some tricky issues around timestamp handling (e.g. in
>> particular in conjunction with deletes). As we need to migrate HBase
>> stored data (for other reasons) shortly we are wondering, if our
>> approach has some long-term drawbacks that we should pay attention to
>> now and possibly re-design our timestamp handling as well.
>>
>> So my question is:
>>
>> * Is there problematic experience with using HBase timestamps as time
>> dimension of your data (assuming it has some natural time-based versioning)?
>>
>> * Is it generally better to model time-based versioning of data within
>> the data structure itself (e.g. in the row key) and why?
>>
>> * In case you used HBase timestamps similar to the way we use them,
>> feedback on how that worked is welcome as well!
>>
>> Thanks,
>> Henning
>>
> The opinions expressed here are mine, while they may reflect a cognitive thought, that
is purely accidental.
> Use at your own risk.
> Michael Segel
> michael_segel (AT) hotmail.com
>
>
>
>
>


-- 
Henning Blohm

*ZFabrik Software KG*

T: 	+49 6227 3984255
F: 	+49 6227 3984254
M: 	+49 1781891820

Lammstrasse 2 69190 Walldorf

henning.blohm@zfabrik.de <mailto:henning.blohm@zfabrik.de>
Linkedin <http://www.linkedin.com/pub/henning-blohm/0/7b5/628>
ZFabrik <http://www.zfabrik.de>
Blog <http://www.z2-environment.net/blog>
Z2-Environment <http://www.z2-environment.eu>
Z2 Wiki <http://redmine.z2-environment.net>


Mime
View raw message