Ignite DataStream vs cache putAll

[daniels <ra...@gmail.com>]

Hi,
Which one TO use  DataStream or cache putAll ?

My cache max-size is  ~50.000 items.

I did some comparison and noticed that in cases of some server nodes and
replicated cace putAll is faster.
Is'nt it?




--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite DataStream vs cache putAll

[daniels <ra...@gmail.com>]

Thank you dear Denis,I will follow to your suggestions. 



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite DataStream vs cache putAll

[Denis Magda <dm...@apache.org>]

I would suggest the data streamer for initial data preloading or when you stream data continuously into the cluster.

Personally, I’ve never seen putAll to beat the data streamer. Check that you have allowOverwrite set to true. Plus you can tweak another parameters.

—
Denis

> On Nov 28, 2017, at 5:37 AM, daniels <ra...@gmail.com> wrote:
> 
> Thank you for response.
> 
> So.in my case(50.000 max size and no need server side processing ) is beter
> to use putAll?
> In other words, is  50.000 very big ammount data?
> 
> 
> 
> --
> Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite DataStream vs cache putAll

[daniels <ra...@gmail.com>]

Thank you for response.

So.in my case(50.000 max size and no need server side processing ) is beter
to use putAll?
In other words, is  50.000 very big ammount data?



--
Sent from: http://apache-ignite-users.70518.x6.nabble.com/

Re: Ignite DataStream vs cache putAll

[Alexey Kukushkin <ku...@gmail.com>]

Hi,

putAll just sends all data right away.

DataStreamer locally builds batches of 512 entries (configurable) and does
an equivalent of "putAll" for the batches as they become full. DataStreamer
also retries "putAll" if it fails. DataStreamer allows you to configure
server-side processing of added data. DataStreamer blocks addData if remote
nodes need more time to process data.

Thus, if you have not a very big amount of static data then putAll shall be
faster. If you have huge data set or need a server side processing then
DataStreamer is a better option.