You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@ignite.apache.org by waterg <je...@gmail.com> on 2017/05/04 00:36:26 UTC
Query search performance and scalability
I have a program that issues SQLQuery on colA, colA has a single column
index.
The cache has about 360K entries, and 298K unique values in colA.
I ran the query 1M times with different values.
If the cache only has cache key, here's the results I ran with three senario
With 1 server node: average search time is 35K nanosecs
With 1 server – 1 client: average search time is 128K nanosecs
With 2 server – 1 client: average search time is 276K nanosecs
If the cache has Affinity Key, here's the results I ran with three senario
With 1 server node: average search time is 59K nanosecs
With 1 server – 1 client: average search time is 148K nanosecs
With 2 server – 1 client: average search time is 287K nanosecs
Could anyone advise why the search speed degrades as more nodes are added?
Thank you!
--
View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Query-search-performance-and-scalability-tp12401.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.
Re: Query search performance and scalability
Posted by Andrey Mashenkov <an...@gmail.com>.
Hi,
Network roundtrip costs 100-200mcs. As you can see error is very high,
lower and higher value differ in more than 2 times.
So, query runinng in 280mcs looks ok.
Netwok latencies often unpredictable, I would think queries per second
metric can give more clear picture than comparing of micro or nano seconds.
On Fri, May 5, 2017 at 8:38 PM, Jessie Lin <je...@gmail.com>
wrote:
> Hi Alex, the ping between the two nodes are mostly within 0.1-0.2 ms. I
> did see once in a while couple ms. Is this the normal range you'd expect?
>
> Jessie
>
>
>
> On Thu, May 4, 2017 at 11:37 PM, afedotov <al...@gmail.com>
> wrote:
>
>> Hi Jessie.
>>
>> What is the ping between physical nodes?
>> As well, please try adding two more nodes and share the results.
>> I think in your case the query is pretty fast initially, so adding more
>> nodes just add to the total time because of more network messages being
>> sent. In real life scenarios when executing a query takes several times
>> more than network overhead you will gain from scaling the cluster.
>>
>>
>>
>> Kind regards,
>> Alex
>>
>>
>> 5 мая 2017 г. 8:04 AM пользователь "waterg [via Apache Ignite Users]" <[hidden
>> email] <http:///user/SendEmail.jtp?type=node&node=12448&i=0>> написал:
>>
>> Here's the cache configuration.
>> We were expecting to see positive scalability.
>> What would be reasons to prevent from that?
>>
>> Btw, the cache has 300K entries in total.
>> Thank you!
>>
>> <property name="cacheConfiguration">
>>
>> <list>
>> <bean class="org.apache.ignite.configuration.CacheConfiguration">
>> <property name="name" value="sampleCache"></property>
>> <!-- Enable readThrough-->
>> <property name="readThrough" value="true"></property>
>> <property name="writeThrough" value="true"></property>
>> <property name="writeBehindEnabled" value="true"></property>
>> <property name="writeBehindFlushSize" value="10240"></property>
>> <property name="writeBehindFlushThreadCount" value="11"></property>
>> <property name="writeBehindBatchSize" value="1024"></property>
>>
>> <!-- Set cacheStoreFactory-->
>> <property name="cacheStoreFactory">
>> <bean class="javax.cache.configuration.FactoryBuilder" factory-method="factoryOf">
>> <constructor-arg value="com.sample.store.SampleStore"></constructor-arg>
>> </bean>
>> </property>
>> <property name="queryEntities">
>> <list>
>> <bean class="org.apache.ignite.cache.QueryEntity">
>> <property name="keyType" value="java.lang.String"></property>
>> <property name="valueType" value="com.sample.model.SampleModel"></property>
>> <property name="fields">
>> <map>
>> <entry key="col1" value="java.lang.Long"></entry>
>> <entry key="col2" value="java.lang.String"></entry>
>> <entry key="col3" value="java.util.String"></entry>
>> </map>
>> </property>
>> <property name="indexes">
>> <list>
>> <!-- Single field (aka. column) index -->
>> <bean class="org.apache.ignite.cache.QueryIndex">
>> <constructor-arg value="col1"/>
>> </bean>
>> <bean class="org.apache.ignite.cache.QueryIndex">
>> <constructor-arg value="col2"/>
>> </bean>
>> </list>
>> </property>
>> </bean>
>> </list>
>> </property>
>>
>> </bean>
>> </list>
>> </property>
>>
>>
>> On Thu, May 4, 2017 at 9:35 AM, Jessie Lin <[hidden email]
>> <http:///user/SendEmail.jtp?type=node&node=12447&i=0>> wrote:
>>
>>> Alex, thanks for reply.
>>> The query is fairly simple, colA =?
>>> colA is String.
>>> For 1 server 1 client scenario, they are the same server.
>>> For 2 server 1 client scenario, one server and client are on one server,
>>> the other server node is on another server.
>>> Working on a simplified code to reproduce the problem to share with you
>>>
>>> Jessie
>>>
>>>
>>> On Wed, May 3, 2017 at 11:14 PM, afedotov <[hidden email]
>>> <http:///user/SendEmail.jtp?type=node&node=12447&i=1>> wrote:
>>>
>>>> That should be because of the overhead of network roundtrips between
>>>> the client and the servers. Are servers and client located on the same
>>>> physical machine?
>>>> Please provide the query and cache config.
>>>>
>>>> Kind regards,
>>>> Alex
>>>>
>>>> 4 мая 2017 г. 3:36 AM пользователь "waterg [via Apache Ignite Users]" <[hidden
>>>> email] <http:///user/SendEmail.jtp?type=node&node=12407&i=0>> написал:
>>>>
>>>> I have a program that issues SQLQuery on colA, colA has a single column
>>>> index.
>>>> The cache has about 360K entries, and 298K unique values in colA.
>>>> I ran the query 1M times with different values.
>>>>
>>>> If the cache only has cache key, here's the results I ran with three
>>>> senario
>>>> With 1 server node: average search time is 35K nanosecs
>>>> With 1 server – 1 client: average search time is 128K nanosecs
>>>> With 2 server – 1 client: average search time is 276K nanosecs
>>>>
>>>> If the cache has Affinity Key, here's the results I ran with three
>>>> senario
>>>> With 1 server node: average search time is 59K nanosecs
>>>> With 1 server – 1 client: average search time is 148K nanosecs
>>>> With 2 server – 1 client: average search time is 287K nanosecs
>>>>
>>>> Could anyone advise why the search speed degrades as more nodes are
>>>> added?
>>>> Thank you!
>>>>
>>>> ------------------------------
>>>> If you reply to this email, your message will be added to the
>>>> discussion below:
>>>> http://apache-ignite-users.70518.x6.nabble.com/Query-search-
>>>> performance-and-scalability-tp12401.html
>>>> To start a new topic under Apache Ignite Users, email [hidden email]
>>>> <http:///user/SendEmail.jtp?type=node&node=12407&i=1>
>>>> To unsubscribe from Apache Ignite Users, click here.
>>>> NAML
>>>> <http://apache-ignite-users.70518.x6.nabble.com/template/NamlServlet.jtp?macro=macro_viewer&id=instant_html%21nabble%3Aemail.naml&base=nabble.naml.namespaces.BasicNamespace-nabble.view.web.template.NabbleNamespace-nabble.view.web.template.NodeNamespace&breadcrumbs=notify_subscribers%21nabble%3Aemail.naml-instant_emails%21nabble%3Aemail.naml-send_instant_email%21nabble%3Aemail.naml>
>>>>
>>>>
>>>>
>>>> ------------------------------
>>>> View this message in context: Re: Query search performance and
>>>> scalability
>>>> <http://apache-ignite-users.70518.x6.nabble.com/Query-search-performance-and-scalability-tp12401p12407.html>
>>>>
>>>> Sent from the Apache Ignite Users mailing list archive
>>>> <http://apache-ignite-users.70518.x6.nabble.com/> at Nabble.com.
>>>>
>>>
>>>
>>
>>
>> ------------------------------
>> If you reply to this email, your message will be added to the discussion
>> below:
>> http://apache-ignite-users.70518.x6.nabble.com/Query-search-
>> performance-and-scalability-tp12401p12447.html
>> To start a new topic under Apache Ignite Users, email [hidden email]
>> <http:///user/SendEmail.jtp?type=node&node=12448&i=1>
>> To unsubscribe from Apache Ignite Users, click here.
>> NAML
>> <http://apache-ignite-users.70518.x6.nabble.com/template/NamlServlet.jtp?macro=macro_viewer&id=instant_html%21nabble%3Aemail.naml&base=nabble.naml.namespaces.BasicNamespace-nabble.view.web.template.NabbleNamespace-nabble.view.web.template.NodeNamespace&breadcrumbs=notify_subscribers%21nabble%3Aemail.naml-instant_emails%21nabble%3Aemail.naml-send_instant_email%21nabble%3Aemail.naml>
>>
>>
>>
>> ------------------------------
>> View this message in context: Re: Query search performance and
>> scalability
>> <http://apache-ignite-users.70518.x6.nabble.com/Query-search-performance-and-scalability-tp12401p12448.html>
>> Sent from the Apache Ignite Users mailing list archive
>> <http://apache-ignite-users.70518.x6.nabble.com/> at Nabble.com.
>>
>
>
--
Best regards,
Andrey V. Mashenkov
Re: Query search performance and scalability
Posted by Jessie Lin <je...@gmail.com>.
Hi Alex, the ping between the two nodes are mostly within 0.1-0.2 ms. I did
see once in a while couple ms. Is this the normal range you'd expect?
Jessie
On Thu, May 4, 2017 at 11:37 PM, afedotov <al...@gmail.com>
wrote:
> Hi Jessie.
>
> What is the ping between physical nodes?
> As well, please try adding two more nodes and share the results.
> I think in your case the query is pretty fast initially, so adding more
> nodes just add to the total time because of more network messages being
> sent. In real life scenarios when executing a query takes several times
> more than network overhead you will gain from scaling the cluster.
>
>
>
> Kind regards,
> Alex
>
>
> 5 мая 2017 г. 8:04 AM пользователь "waterg [via Apache Ignite Users]" <[hidden
> email] <http:///user/SendEmail.jtp?type=node&node=12448&i=0>> написал:
>
> Here's the cache configuration.
> We were expecting to see positive scalability.
> What would be reasons to prevent from that?
>
> Btw, the cache has 300K entries in total.
> Thank you!
>
> <property name="cacheConfiguration">
>
> <list>
> <bean class="org.apache.ignite.configuration.CacheConfiguration">
> <property name="name" value="sampleCache"></property>
> <!-- Enable readThrough-->
> <property name="readThrough" value="true"></property>
> <property name="writeThrough" value="true"></property>
> <property name="writeBehindEnabled" value="true"></property>
> <property name="writeBehindFlushSize" value="10240"></property>
> <property name="writeBehindFlushThreadCount" value="11"></property>
> <property name="writeBehindBatchSize" value="1024"></property>
>
> <!-- Set cacheStoreFactory-->
> <property name="cacheStoreFactory">
> <bean class="javax.cache.configuration.FactoryBuilder" factory-method="factoryOf">
> <constructor-arg value="com.sample.store.SampleStore"></constructor-arg>
> </bean>
> </property>
> <property name="queryEntities">
> <list>
> <bean class="org.apache.ignite.cache.QueryEntity">
> <property name="keyType" value="java.lang.String"></property>
> <property name="valueType" value="com.sample.model.SampleModel"></property>
> <property name="fields">
> <map>
> <entry key="col1" value="java.lang.Long"></entry>
> <entry key="col2" value="java.lang.String"></entry>
> <entry key="col3" value="java.util.String"></entry>
> </map>
> </property>
> <property name="indexes">
> <list>
> <!-- Single field (aka. column) index -->
> <bean class="org.apache.ignite.cache.QueryIndex">
> <constructor-arg value="col1"/>
> </bean>
> <bean class="org.apache.ignite.cache.QueryIndex">
> <constructor-arg value="col2"/>
> </bean>
> </list>
> </property>
> </bean>
> </list>
> </property>
>
> </bean>
> </list>
> </property>
>
>
> On Thu, May 4, 2017 at 9:35 AM, Jessie Lin <[hidden email]
> <http:///user/SendEmail.jtp?type=node&node=12447&i=0>> wrote:
>
>> Alex, thanks for reply.
>> The query is fairly simple, colA =?
>> colA is String.
>> For 1 server 1 client scenario, they are the same server.
>> For 2 server 1 client scenario, one server and client are on one server,
>> the other server node is on another server.
>> Working on a simplified code to reproduce the problem to share with you
>>
>> Jessie
>>
>>
>> On Wed, May 3, 2017 at 11:14 PM, afedotov <[hidden email]
>> <http:///user/SendEmail.jtp?type=node&node=12447&i=1>> wrote:
>>
>>> That should be because of the overhead of network roundtrips between the
>>> client and the servers. Are servers and client located on the same physical
>>> machine?
>>> Please provide the query and cache config.
>>>
>>> Kind regards,
>>> Alex
>>>
>>> 4 мая 2017 г. 3:36 AM пользователь "waterg [via Apache Ignite Users]" <[hidden
>>> email] <http:///user/SendEmail.jtp?type=node&node=12407&i=0>> написал:
>>>
>>> I have a program that issues SQLQuery on colA, colA has a single column
>>> index.
>>> The cache has about 360K entries, and 298K unique values in colA.
>>> I ran the query 1M times with different values.
>>>
>>> If the cache only has cache key, here's the results I ran with three
>>> senario
>>> With 1 server node: average search time is 35K nanosecs
>>> With 1 server – 1 client: average search time is 128K nanosecs
>>> With 2 server – 1 client: average search time is 276K nanosecs
>>>
>>> If the cache has Affinity Key, here's the results I ran with three
>>> senario
>>> With 1 server node: average search time is 59K nanosecs
>>> With 1 server – 1 client: average search time is 148K nanosecs
>>> With 2 server – 1 client: average search time is 287K nanosecs
>>>
>>> Could anyone advise why the search speed degrades as more nodes are
>>> added?
>>> Thank you!
>>>
>>> ------------------------------
>>> If you reply to this email, your message will be added to the discussion
>>> below:
>>> http://apache-ignite-users.70518.x6.nabble.com/Query-search-
>>> performance-and-scalability-tp12401.html
>>> To start a new topic under Apache Ignite Users, email [hidden email]
>>> <http:///user/SendEmail.jtp?type=node&node=12407&i=1>
>>> To unsubscribe from Apache Ignite Users, click here.
>>> NAML
>>> <http://apache-ignite-users.70518.x6.nabble.com/template/NamlServlet.jtp?macro=macro_viewer&id=instant_html%21nabble%3Aemail.naml&base=nabble.naml.namespaces.BasicNamespace-nabble.view.web.template.NabbleNamespace-nabble.view.web.template.NodeNamespace&breadcrumbs=notify_subscribers%21nabble%3Aemail.naml-instant_emails%21nabble%3Aemail.naml-send_instant_email%21nabble%3Aemail.naml>
>>>
>>>
>>>
>>> ------------------------------
>>> View this message in context: Re: Query search performance and
>>> scalability
>>> <http://apache-ignite-users.70518.x6.nabble.com/Query-search-performance-and-scalability-tp12401p12407.html>
>>>
>>> Sent from the Apache Ignite Users mailing list archive
>>> <http://apache-ignite-users.70518.x6.nabble.com/> at Nabble.com.
>>>
>>
>>
>
>
> ------------------------------
> If you reply to this email, your message will be added to the discussion
> below:
> http://apache-ignite-users.70518.x6.nabble.com/Query-search-
> performance-and-scalability-tp12401p12447.html
> To start a new topic under Apache Ignite Users, email [hidden email]
> <http:///user/SendEmail.jtp?type=node&node=12448&i=1>
> To unsubscribe from Apache Ignite Users, click here.
> NAML
> <http://apache-ignite-users.70518.x6.nabble.com/template/NamlServlet.jtp?macro=macro_viewer&id=instant_html%21nabble%3Aemail.naml&base=nabble.naml.namespaces.BasicNamespace-nabble.view.web.template.NabbleNamespace-nabble.view.web.template.NodeNamespace&breadcrumbs=notify_subscribers%21nabble%3Aemail.naml-instant_emails%21nabble%3Aemail.naml-send_instant_email%21nabble%3Aemail.naml>
>
>
>
> ------------------------------
> View this message in context: Re: Query search performance and scalability
> <http://apache-ignite-users.70518.x6.nabble.com/Query-search-performance-and-scalability-tp12401p12448.html>
> Sent from the Apache Ignite Users mailing list archive
> <http://apache-ignite-users.70518.x6.nabble.com/> at Nabble.com.
>
Re: Query search performance and scalability
Posted by afedotov <al...@gmail.com>.
Hi Jessie.
What is the ping between physical nodes?
As well, please try adding two more nodes and share the results.
I think in your case the query is pretty fast initially, so adding more
nodes just add to the total time because of more network messages being
sent. In real life scenarios when executing a query takes several times
more than network overhead you will gain from scaling the cluster.
Kind regards,
Alex
5 мая 2017 г. 8:04 AM пользователь "waterg [via Apache Ignite Users]" <
ml+s70518n12447h91@n6.nabble.com> написал:
Here's the cache configuration.
We were expecting to see positive scalability.
What would be reasons to prevent from that?
Btw, the cache has 300K entries in total.
Thank you!
<property name="cacheConfiguration">
<list>
<bean class="org.apache.ignite.configuration.CacheConfiguration">
<property name="name" value="sampleCache"></property>
<!-- Enable readThrough-->
<property name="readThrough" value="true"></property>
<property name="writeThrough" value="true"></property>
<property name="writeBehindEnabled" value="true"></property>
<property name="writeBehindFlushSize" value="10240"></property>
<property name="writeBehindFlushThreadCount" value="11"></property>
<property name="writeBehindBatchSize" value="1024"></property>
<!-- Set cacheStoreFactory-->
<property name="cacheStoreFactory">
<bean class="javax.cache.configuration.FactoryBuilder"
factory-method="factoryOf">
<constructor-arg
value="com.sample.store.SampleStore"></constructor-arg>
</bean>
</property>
<property name="queryEntities">
<list>
<bean class="org.apache.ignite.cache.QueryEntity">
<property name="keyType"
value="java.lang.String"></property>
<property name="valueType"
value="com.sample.model.SampleModel"></property>
<property name="fields">
<map>
<entry key="col1" value="java.lang.Long"></entry>
<entry key="col2" value="java.lang.String"></entry>
<entry key="col3" value="java.util.String"></entry>
</map>
</property>
<property name="indexes">
<list>
<!-- Single field (aka. column) index -->
<bean class="org.apache.ignite.cache.QueryIndex">
<constructor-arg value="col1"/>
</bean>
<bean class="org.apache.ignite.cache.QueryIndex">
<constructor-arg value="col2"/>
</bean>
</list>
</property>
</bean>
</list>
</property>
</bean>
</list>
</property>
On Thu, May 4, 2017 at 9:35 AM, Jessie Lin <[hidden email]
<http:///user/SendEmail.jtp?type=node&node=12447&i=0>> wrote:
> Alex, thanks for reply.
> The query is fairly simple, colA =?
> colA is String.
> For 1 server 1 client scenario, they are the same server.
> For 2 server 1 client scenario, one server and client are on one server,
> the other server node is on another server.
> Working on a simplified code to reproduce the problem to share with you
>
> Jessie
>
>
> On Wed, May 3, 2017 at 11:14 PM, afedotov <[hidden email]
> <http:///user/SendEmail.jtp?type=node&node=12447&i=1>> wrote:
>
>> That should be because of the overhead of network roundtrips between the
>> client and the servers. Are servers and client located on the same physical
>> machine?
>> Please provide the query and cache config.
>>
>> Kind regards,
>> Alex
>>
>> 4 мая 2017 г. 3:36 AM пользователь "waterg [via Apache Ignite Users]" <[hidden
>> email] <http:///user/SendEmail.jtp?type=node&node=12407&i=0>> написал:
>>
>> I have a program that issues SQLQuery on colA, colA has a single column
>> index.
>> The cache has about 360K entries, and 298K unique values in colA.
>> I ran the query 1M times with different values.
>>
>> If the cache only has cache key, here's the results I ran with three
>> senario
>> With 1 server node: average search time is 35K nanosecs
>> With 1 server – 1 client: average search time is 128K nanosecs
>> With 2 server – 1 client: average search time is 276K nanosecs
>>
>> If the cache has Affinity Key, here's the results I ran with three
>> senario
>> With 1 server node: average search time is 59K nanosecs
>> With 1 server – 1 client: average search time is 148K nanosecs
>> With 2 server – 1 client: average search time is 287K nanosecs
>>
>> Could anyone advise why the search speed degrades as more nodes are
>> added?
>> Thank you!
>>
>> ------------------------------
>> If you reply to this email, your message will be added to the discussion
>> below:
>> http://apache-ignite-users.70518.x6.nabble.com/Query-search-
>> performance-and-scalability-tp12401.html
>> To start a new topic under Apache Ignite Users, email [hidden email]
>> <http:///user/SendEmail.jtp?type=node&node=12407&i=1>
>> To unsubscribe from Apache Ignite Users, click here.
>> NAML
>> <http://apache-ignite-users.70518.x6.nabble.com/template/NamlServlet.jtp?macro=macro_viewer&id=instant_html%21nabble%3Aemail.naml&base=nabble.naml.namespaces.BasicNamespace-nabble.view.web.template.NabbleNamespace-nabble.view.web.template.NodeNamespace&breadcrumbs=notify_subscribers%21nabble%3Aemail.naml-instant_emails%21nabble%3Aemail.naml-send_instant_email%21nabble%3Aemail.naml>
>>
>>
>>
>> ------------------------------
>> View this message in context: Re: Query search performance and
>> scalability
>> <http://apache-ignite-users.70518.x6.nabble.com/Query-search-performance-and-scalability-tp12401p12407.html>
>>
>> Sent from the Apache Ignite Users mailing list archive
>> <http://apache-ignite-users.70518.x6.nabble.com/> at Nabble.com.
>>
>
>
------------------------------
If you reply to this email, your message will be added to the discussion
below:
http://apache-ignite-users.70518.x6.nabble.com/Query-search-performance-and-
scalability-tp12401p12447.html
To start a new topic under Apache Ignite Users, email
ml+s70518n1h65@n6.nabble.com
To unsubscribe from Apache Ignite Users, click here
<http://apache-ignite-users.70518.x6.nabble.com/template/NamlServlet.jtp?macro=unsubscribe_by_code&node=1&code=YWxleGFuZGVyLmZlZG90b2ZmQGdtYWlsLmNvbXwxfC0xMzYxNTU0NTg=>
.
NAML
<http://apache-ignite-users.70518.x6.nabble.com/template/NamlServlet.jtp?macro=macro_viewer&id=instant_html%21nabble%3Aemail.naml&base=nabble.naml.namespaces.BasicNamespace-nabble.view.web.template.NabbleNamespace-nabble.view.web.template.NodeNamespace&breadcrumbs=notify_subscribers%21nabble%3Aemail.naml-instant_emails%21nabble%3Aemail.naml-send_instant_email%21nabble%3Aemail.naml>
--
View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Query-search-performance-and-scalability-tp12401p12448.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.
Re: Query search performance and scalability
Posted by Jessie Lin <je...@gmail.com>.
Here's the cache configuration.
We were expecting to see positive scalability.
What would be reasons to prevent from that?
Btw, the cache has 300K entries in total.
Thank you!
<property name="cacheConfiguration">
<list>
<bean class="org.apache.ignite.configuration.CacheConfiguration">
<property name="name" value="sampleCache"></property>
<!-- Enable readThrough-->
<property name="readThrough" value="true"></property>
<property name="writeThrough" value="true"></property>
<property name="writeBehindEnabled" value="true"></property>
<property name="writeBehindFlushSize" value="10240"></property>
<property name="writeBehindFlushThreadCount" value="11"></property>
<property name="writeBehindBatchSize" value="1024"></property>
<!-- Set cacheStoreFactory-->
<property name="cacheStoreFactory">
<bean class="javax.cache.configuration.FactoryBuilder"
factory-method="factoryOf">
<constructor-arg
value="com.sample.store.SampleStore"></constructor-arg>
</bean>
</property>
<property name="queryEntities">
<list>
<bean class="org.apache.ignite.cache.QueryEntity">
<property name="keyType"
value="java.lang.String"></property>
<property name="valueType"
value="com.sample.model.SampleModel"></property>
<property name="fields">
<map>
<entry key="col1" value="java.lang.Long"></entry>
<entry key="col2" value="java.lang.String"></entry>
<entry key="col3" value="java.util.String"></entry>
</map>
</property>
<property name="indexes">
<list>
<!-- Single field (aka. column) index -->
<bean class="org.apache.ignite.cache.QueryIndex">
<constructor-arg value="col1"/>
</bean>
<bean class="org.apache.ignite.cache.QueryIndex">
<constructor-arg value="col2"/>
</bean>
</list>
</property>
</bean>
</list>
</property>
</bean>
</list>
</property>
On Thu, May 4, 2017 at 9:35 AM, Jessie Lin <je...@gmail.com>
wrote:
> Alex, thanks for reply.
> The query is fairly simple, colA =?
> colA is String.
> For 1 server 1 client scenario, they are the same server.
> For 2 server 1 client scenario, one server and client are on one server,
> the other server node is on another server.
> Working on a simplified code to reproduce the problem to share with you
>
> Jessie
>
>
> On Wed, May 3, 2017 at 11:14 PM, afedotov <al...@gmail.com>
> wrote:
>
>> That should be because of the overhead of network roundtrips between the
>> client and the servers. Are servers and client located on the same physical
>> machine?
>> Please provide the query and cache config.
>>
>> Kind regards,
>> Alex
>>
>> 4 мая 2017 г. 3:36 AM пользователь "waterg [via Apache Ignite Users]" <[hidden
>> email] <http:///user/SendEmail.jtp?type=node&node=12407&i=0>> написал:
>>
>> I have a program that issues SQLQuery on colA, colA has a single column
>> index.
>> The cache has about 360K entries, and 298K unique values in colA.
>> I ran the query 1M times with different values.
>>
>> If the cache only has cache key, here's the results I ran with three
>> senario
>> With 1 server node: average search time is 35K nanosecs
>> With 1 server – 1 client: average search time is 128K nanosecs
>> With 2 server – 1 client: average search time is 276K nanosecs
>>
>> If the cache has Affinity Key, here's the results I ran with three
>> senario
>> With 1 server node: average search time is 59K nanosecs
>> With 1 server – 1 client: average search time is 148K nanosecs
>> With 2 server – 1 client: average search time is 287K nanosecs
>>
>> Could anyone advise why the search speed degrades as more nodes are
>> added?
>> Thank you!
>>
>> ------------------------------
>> If you reply to this email, your message will be added to the discussion
>> below:
>> http://apache-ignite-users.70518.x6.nabble.com/Query-search-
>> performance-and-scalability-tp12401.html
>> To start a new topic under Apache Ignite Users, email [hidden email]
>> <http:///user/SendEmail.jtp?type=node&node=12407&i=1>
>> To unsubscribe from Apache Ignite Users, click here.
>> NAML
>> <http://apache-ignite-users.70518.x6.nabble.com/template/NamlServlet.jtp?macro=macro_viewer&id=instant_html%21nabble%3Aemail.naml&base=nabble.naml.namespaces.BasicNamespace-nabble.view.web.template.NabbleNamespace-nabble.view.web.template.NodeNamespace&breadcrumbs=notify_subscribers%21nabble%3Aemail.naml-instant_emails%21nabble%3Aemail.naml-send_instant_email%21nabble%3Aemail.naml>
>>
>>
>>
>> ------------------------------
>> View this message in context: Re: Query search performance and
>> scalability
>> <http://apache-ignite-users.70518.x6.nabble.com/Query-search-performance-and-scalability-tp12401p12407.html>
>>
>> Sent from the Apache Ignite Users mailing list archive
>> <http://apache-ignite-users.70518.x6.nabble.com/> at Nabble.com.
>>
>
>
Re: Query search performance and scalability
Posted by Jessie Lin <je...@gmail.com>.
Alex, thanks for reply.
The query is fairly simple, colA =?
colA is String.
For 1 server 1 client scenario, they are the same server.
For 2 server 1 client scenario, one server and client are on one server,
the other server node is on another server.
Working on a simplified code to reproduce the problem to share with you
Jessie
On Wed, May 3, 2017 at 11:14 PM, afedotov <al...@gmail.com>
wrote:
> That should be because of the overhead of network roundtrips between the
> client and the servers. Are servers and client located on the same physical
> machine?
> Please provide the query and cache config.
>
> Kind regards,
> Alex
>
> 4 мая 2017 г. 3:36 AM пользователь "waterg [via Apache Ignite Users]" <[hidden
> email] <http:///user/SendEmail.jtp?type=node&node=12407&i=0>> написал:
>
> I have a program that issues SQLQuery on colA, colA has a single column
> index.
> The cache has about 360K entries, and 298K unique values in colA.
> I ran the query 1M times with different values.
>
> If the cache only has cache key, here's the results I ran with three
> senario
> With 1 server node: average search time is 35K nanosecs
> With 1 server – 1 client: average search time is 128K nanosecs
> With 2 server – 1 client: average search time is 276K nanosecs
>
> If the cache has Affinity Key, here's the results I ran with three senario
> With 1 server node: average search time is 59K nanosecs
> With 1 server – 1 client: average search time is 148K nanosecs
> With 2 server – 1 client: average search time is 287K nanosecs
>
> Could anyone advise why the search speed degrades as more nodes are added?
> Thank you!
>
> ------------------------------
> If you reply to this email, your message will be added to the discussion
> below:
> http://apache-ignite-users.70518.x6.nabble.com/Query-search-
> performance-and-scalability-tp12401.html
> To start a new topic under Apache Ignite Users, email [hidden email]
> <http:///user/SendEmail.jtp?type=node&node=12407&i=1>
> To unsubscribe from Apache Ignite Users, click here.
> NAML
> <http://apache-ignite-users.70518.x6.nabble.com/template/NamlServlet.jtp?macro=macro_viewer&id=instant_html%21nabble%3Aemail.naml&base=nabble.naml.namespaces.BasicNamespace-nabble.view.web.template.NabbleNamespace-nabble.view.web.template.NodeNamespace&breadcrumbs=notify_subscribers%21nabble%3Aemail.naml-instant_emails%21nabble%3Aemail.naml-send_instant_email%21nabble%3Aemail.naml>
>
>
>
> ------------------------------
> View this message in context: Re: Query search performance and scalability
> <http://apache-ignite-users.70518.x6.nabble.com/Query-search-performance-and-scalability-tp12401p12407.html>
>
> Sent from the Apache Ignite Users mailing list archive
> <http://apache-ignite-users.70518.x6.nabble.com/> at Nabble.com.
>
Re: Query search performance and scalability
Posted by afedotov <al...@gmail.com>.
That should be because of the overhead of network roundtrips between the
client and the servers. Are servers and client located on the same physical
machine?
Please provide the query and cache config.
Kind regards,
Alex
4 мая 2017 г. 3:36 AM пользователь "waterg [via Apache Ignite Users]" <
ml+s70518n12401h65@n6.nabble.com> написал:
I have a program that issues SQLQuery on colA, colA has a single column
index.
The cache has about 360K entries, and 298K unique values in colA.
I ran the query 1M times with different values.
If the cache only has cache key, here's the results I ran with three
senario
With 1 server node: average search time is 35K nanosecs
With 1 server – 1 client: average search time is 128K nanosecs
With 2 server – 1 client: average search time is 276K nanosecs
If the cache has Affinity Key, here's the results I ran with three senario
With 1 server node: average search time is 59K nanosecs
With 1 server – 1 client: average search time is 148K nanosecs
With 2 server – 1 client: average search time is 287K nanosecs
Could anyone advise why the search speed degrades as more nodes are added?
Thank you!
------------------------------
If you reply to this email, your message will be added to the discussion
below:
http://apache-ignite-users.70518.x6.nabble.com/Query-search-performance-and-
scalability-tp12401.html
To start a new topic under Apache Ignite Users, email
ml+s70518n1h65@n6.nabble.com
To unsubscribe from Apache Ignite Users, click here
<http://apache-ignite-users.70518.x6.nabble.com/template/NamlServlet.jtp?macro=unsubscribe_by_code&node=1&code=YWxleGFuZGVyLmZlZG90b2ZmQGdtYWlsLmNvbXwxfC0xMzYxNTU0NTg=>
.
NAML
<http://apache-ignite-users.70518.x6.nabble.com/template/NamlServlet.jtp?macro=macro_viewer&id=instant_html%21nabble%3Aemail.naml&base=nabble.naml.namespaces.BasicNamespace-nabble.view.web.template.NabbleNamespace-nabble.view.web.template.NodeNamespace&breadcrumbs=notify_subscribers%21nabble%3Aemail.naml-instant_emails%21nabble%3Aemail.naml-send_instant_email%21nabble%3Aemail.naml>
--
View this message in context: http://apache-ignite-users.70518.x6.nabble.com/Query-search-performance-and-scalability-tp12401p12407.html
Sent from the Apache Ignite Users mailing list archive at Nabble.com.