curator-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Cameron McKenzie <mckenzie....@gmail.com>
Subject Re: tree cache load time
Date Wed, 16 Sep 2015 00:15:37 GMT
I don't think you'll be able to extend the existing TreeCache to handle
multi transactions. It would need to be a modification within Curator.

How much data are you actually caching? Jordan's suggestion of using the
multi transactions is certainly a good one and would provide some
performance benefits, but I have no idea how much.

On Wed, Sep 16, 2015 at 10:06 AM, Vikrant Singh <vikrant.subscribe@gmail.com
> wrote:

> So does this mean I should try to modify tree cache to do multi
> transaction? I was wondering if this is something I can do on top of stock
> tree cache? if yes .. any pointers on how?
>
> On Tue, Sep 15, 2015 at 4:59 PM, Cameron McKenzie <mckenzie.cam@gmail.com>
> wrote:
>
>> How would you do multiple levels? Maybe I'm missing something, but don't
>> you need to know the names of the child nodes for a level before you can
>> query the level below that?
>>
>> On Wed, Sep 16, 2015 at 9:58 AM, Jordan Zimmerman <
>> jordan@jordanzimmerman.com> wrote:
>>
>>> You could even do multiple levels. Of course, you’d run into the 1MB
>>> limit at some point. But, maybe if TreeCache knows it needs to do a certain
>>> amount of work it could batch it via multi-transaction.
>>>
>>>
>>> On September 15, 2015 at 6:57:36 PM, Cameron McKenzie (
>>> mckenzie.cam@gmail.com) wrote:
>>>
>>> So you could execute the all the required getChildren calls for a
>>> particular level of the tree in a single call to ZK?
>>>
>>> On Wed, Sep 16, 2015 at 9:54 AM, Jordan Zimmerman <
>>> jordan@jordanzimmerman.com> wrote:
>>>
>>>> However, there is an optimization opportunity by using ZK’s
>>>> multi-transactions. Maybe we can explore this.
>>>>
>>>> -Jordan
>>>>
>>>>
>>>>
>>>> On September 15, 2015 at 6:44:46 PM, Cameron McKenzie (
>>>> mckenzie.cam@gmail.com) wrote:
>>>>
>>>> It would be largely dependent upon how much data you're caching. The
>>>> tree cache needs to recursively query ZK to populate the cache.
>>>> cheers
>>>>
>>>> On Wed, Sep 16, 2015 at 9:30 AM, Vikrant Singh <
>>>> vikrant.subscribe@gmail.com> wrote:
>>>>
>>>>> Hello All,
>>>>> I am using tree cache for one of my project. Issue I am facing is with
>>>>> the time it is taking for initialize itself. It is taking around 300ms
to
>>>>> 3-4 s to load itself.
>>>>> This is how I am initializing the client and cache
>>>>>
>>>>>
>>>>> * private val client = CuratorFrameworkFactory.newClient(hostList, new
>>>>> ExponentialBackoffRetry(ZookeeperCache.RetryInterval.toMillis.toInt,ZookeeperCache.RetryCount))*
>>>>>
>>>>>
>>>>> *val cache = TreeCache.newBuilder(client, path).build()*
>>>>>
>>>>>
>>>>> post this I register for  a handler and then wait for
>>>>> "TreeCacheEvent.Type.INITIALIZED" event.  It take on avg  300ms to 3-4
s.
>>>>> IS there something I can do to improve this performance?
>>>>> Thanks,
>>>>> Vikrant
>>>>>
>>>>>
>>>>>
>>>>
>>>
>>
>

Mime
View raw message