Azure Search compared to Solr for Sitecore PaaS (Chapter 1: Ingestion)

I’ve been investigating Azure PaaS architectures for Sitecore lately, and I wanted to take a few minutes and summarize some recent findings around the standard Sitecore search providers of Solr and, new for Sitecore PaaS, Azure Search.

To provision Azure PaaS Sitecore environments, I used a variant of the ARM Template approach outlined in this blog.  For simplicity, I evaluated a basic “XP-0” which is the name for the Sitecore CM/CD server combined into a single App Service.  This is considered a basic setup for development or testing, but not real production . . . that’s OK for my purposes, however, as I’m interested in comparing the Sitecore search providers to get an idea for relative performance.

The Results

I’ll save the methodology and details for lower in this post, since I’m sure most don’t want to wait for an idea for the results.  The Solr search provider performed faster, no matter the App Service or DB Tier I evaluated in Azure PaaS:

ChartComparison

The chart shows averages to perform the full re-index operation in minutes.  You may want to refer to my earlier post about the lack of HA with Sitecore’s use of Azure Search; rest assured Sitecore is addressing this in a product update soon, but for now it casts a more significant shadow over the 60+ minutes one could spend waiting for the search re-index to complete.

Methodology

In these PaaS trials, I setup the sample site LaunchSitecore.  I performed rebuilds of the sitecore_core_index through the Sitecore Control Panel as my benchmark; I like using this operation as a benchmark since it has over 80,000 documents.  It doesn’t particularly exercise the querying aspects of Sitecore search, though, so I’ll save that dimension for another time.  I’ve got time set aside for JMeter testing that will shed light on this later…

To get the duration the system took to complete the re-index, I queried the PaaS Sitecore logs as described in this Sitecore KB article.  Using results like the following, I took the timestamps since I’ve found the Sitecore UI to be unreliable in reporting duration for index rebuilds.

queries

You can get at this data yourself in App Insights with a query such as this:

traces
| where timestamp > now(-3h)
| where message contains " Crawler [sitecore_core_index]" 
| project timestamp, message
| sort by timestamp desc

Remember, I’ve used the XP0 PaaS ARM Templates which combine CM and CD roles together, so there’s no need for the “where cloud_RoleInstance == ‘CloudRoleBlahBlah'” in the App Insights query.

Methodology – Azure Search

For my Azure Search testing, I experimented with scaling options for Azure Search.  For speedier document ingestion, the guidance from Microsoft says:

“Partitions allow for scaling of document counts as well as faster data ingestion by spanning your index over multiple Azure Search Units”

The trials should perform more quickly with additional Azure Search Partitions, but I found changing this made zero difference.  My instincts tell me the fact Sitecore isn’t using Azure Search Indexers could be a reason scaling Azure Search doesn’t improve performance in my trials.  Sitecore is making REST calls to index documents with Azure Search, which is fine, but possibly not the best fit for high-volume operations.  I haven’t looked in the DLLs, but perhaps there’s other async models one could use in the the Azure Search provider when it comes to full re-indexes?  It could also be that the 80,000 documents in the sitecore_core_index is too small a number to take advantage of Azure Search’s scaling options.  This will be an area for additional research in the future.

Methodology – Solr

To host Solr for this trial, I used a basic Solr VM in the Rackspace cloud.  One benefit to working at Rackspace is easy access to these sorts of resources 🙂  I picked a 4 GB server running Solr 5.5.1.  I used a one Solr core per Sitecore index (1:1 mapping), see my write-up on Solr core organization if you’re not following why this might be relevant.

For my testing with the Solr search provider, Sitecore running  Azure PaaS needed to connect outside Azure, so I selected a location near to Azure US-East where my App Service was hosted.  I had some concerns about outbound data charges, since data leaving Azure will trigger egress bandwidth fees (see this schedule for pricing).  For the few weeks while I collected this data, the outbound data fee totaled less than $40 — and that includes other people using the same Azure account for other experiments.  I estimate around 10% (just $4) is due to my experiments.  Suffice it to say using a Solr environment outside of Azure isn’t a big expense to worry about.  Just the same, running Solr in an Azure VM would certainly be the recommendation for any real Sitecore implementation following this pattern.  For these tests, I chose the Rackspace VM since I already had it handy.

I’d be remiss to not mention the excellent work Sitecore’s Ivan Sharamok has posted to help make Solr truly enterprise ready with Sitecore.  Basic Auth for Solr with Sitecore is important for the architecture I exercised; this post is another gem of Ivan’s worth including here, even if I didn’t make use of it in this specific set of evaluations.  Full disclosure: I worked with Ivan while I was at the Sitecore mothership, so I’m biased that his contributions are valuable, but just because I’m biased doesn’t mean I’m wrong.

Conclusions

I’ll include my chart once again:

ChartComparison

These findings lead me to more questions than answers, so I’m hesitant to make any sweeping generalizations here.  I’m safe declaring Sitecore’s search provider for Solr to be faster than the Azure Search alternative when it comes to full index re-builds, that’s clear by an order of magnitude in some cases.  Know that this is not a judgement about Solr versus Azure Search;  this is about the way Sitecore makes use of these two search technologies out of the box.  The Solr provider for Sitecore is battle-tested and has gone through many years of development; I think the Azure Search provider for Sitecore could be considered a beta at this point, so it’s important to not get ahead of ourselves.

A couple other conclusions could be:

  1. Whether using Solr or Azure Search, there is no improvement to search re-index performance when changing between the S3 to P3 tiers in Azure App Services.
  2. Changing from the S1 to S3 tiers, on the other hand, makes a big perf difference in terms of search re-indexing.
    • Honestly, the S1 tier is almost unusable as the single CPU core and 1.75 GB RAM are way too low for Sitecore; the S3 with 4 cores and 7 GB RAM is much more reasonable to work with.

Next Up

It’s time for me to consider the more fully scaled PaaS options with Sitecore, and I need to exercise the query side of the Sitecore search provider instead of just the indexing side.

Auto-suggest with Solr Facets in Sitecore

Sitecore’s auto-suggest feature for search in the Content Authoring environment is pretty slick, but there is some confusing documentation from Sitecore about how to set it up properly with Solr.  As of today, Sitecore’s documentation on integrating with Solr indicates…

“When you implement Solr with Sitecore you need to enable term support in the Solr search handler.  The term functionality is built into Solr but is disabled by default. To power the dropdowns in the UI you must enable the terms component.

That above documentation will be updated at some point by Sitecore, since it’s no longer the case for the latest version of Sitecore — 8.2 rev. 161221 (Update-2).

In earlier versions of Sitecore, search in the Sitecore Content Editor could make use of the Solr “terms” component to populate suggestions.  This is why this guidance has previously been part of the Solr integration documentation from Sitecore.  Read more about Solr’s use of this auto-suggest through terms at https://cwiki.apache.org/confluence/display/solr/The+Terms+Component.

Sitecore’s strategy of making use of the “terms” component has changed with recently, however.  Sitecore now uses faceting with Solr instead of terms.

To prove this out, I’m going to turn to the Solr logs after I try some queries for content in the Sitecore client.  Refer to this documentation from Sitecore if you’re looking for more context on how to use the search facility — there are a lot of features that are very under-utilized, in my experience.  I’ll specify a clause by typing Updatedby: and then “siteco” to engage the auto-suggest feature:searbhby

Very nice, right?

Under the covers, the Solr logs will reveal something like this . . .

2017-02-17 19:33:07.546 INFO  (qtp33171127-11) [   x:trial_core] o.a.s.c.S.Request [trial_core]  webapp=/solr path=/select params={q=*:*&facet.field=parsedupdatedby_s&facet.prefix=siteco&rows=0&facet=true&version=2.2&facet.sort=true} hits=24626 status=0 QTime=2

. .  . and that can be further debugged by turning it into the URL request powering that auto-suggest response . . .

http://server:port/solr/sitecore_master_index/select?q=*:*&facet.field=parsedupdatedby_s&facet.prefix=siteco&rows=0&facet=true&version=2.2&facet.sort=true

. . . and that would return results like the following:

solrresponse

If instead we tried an author: search in Sitecore, for example, the facet.field would be parsedcreatedby_s instead of parsedupdatedby_s.

I don’t want to go too far down this rabbit hole.  I really just wanted to share that despite what the documentation shows, it’s not necessary to enable the Solr term component on the /select requestHandler in Solr if you’re using the most recent version of Sitecore.  I’ve confirmed with official Sitecore support that this change was tagged as change #444661 and that’s it was incorporated into the product since Sitecore 8.1 update-1 (rev. 151207); the release notes for 8.1 update-1 are vague, but here it is:

Autocomplete for known fields such as language did not work in the Content Editor Search tab using the SOLR provider. The problem was related to the SOLR server configuration. This has been fixed so that Sitecore no longer depends on this configuration. (444661)

Happy faceting to all!

 

Strategies for Sitecore Index Organization into Solr Cores

A few days ago, I shared a graphic I put together to illustrate how Solr can be used to organize Sitecore “indexes” into Solr “cores” — this post has the complete graphic.  I want to elaborate on how one sets Sitecore up to use these two approaches, and dig further into the details.

1:1 Sitecore Index to Solr Core Strategy

To start, here’s a visual showing the typical way Sitecore “indexes” are structured in Solr using a one-to-one (1:1) mapping:

solrseparate

This shows each of the default search indexes defined by Sitecore organized into their own cores defined in Solr.  It’s a 1:1 mapping.  This 1:1 strategy means each index has their own configuration (“conf”) directory in Solr, so seperate stopwords.txt, solrconfig.xml, schema.xml, and so on; it also means each index has their own (“data”) directory in Solr, so separate tlog folders, separate Segment files, etc.

This is the setup one achieves by following the community documentation on setting up Sitecore with Solr; specifically, this quote from that write-up is where you’re doing a lot of the grunt work around setting up distinct Solr cores for each Sitecore index:

“Use the process detailed in Steps 4-7 to create new cores for all the remaining indexes you would like to move to SOLR.”

Since this is the common strategy, I’m not going to go into more details as it’s straight-forward to Sitecore teams.

Kitchen Sink (∞:1 Sitecore Index to Solr Core) Strategy

Here is the comparable graphic showing the ∞:1 strategy of structuring Sitecore indexes in Solr; I like to think of this as the Kitchen Sink container for all Sitecore indexes, since everything goes into that single core just like the kitchen sink:

solrsame

With this approach, a single data and configuration definition is shared by all the Sitecore indexes that reside in Solr.  The advantages are reduced management (setting up the Solr replicationHandler, for example, requires updating 15 solrconfig.xml files in the 1:1 approach, but the Kitchen Sink would require only one solrconfig.xml file to update).  There are significant drawbacks to consider with the Kitchen Sink, however, as you’re sacrificing scaling options specific to each Sitecore index and enforcing a common schema.xml for every index stored in this single core.  There are plenty of reasons not to do this for a production installation of Sitecore, but for a crowded Sitecore environment used for acceptance testing or other use-cases where bullet-proof stability and lots of flexibility when it comes to performance tuning, sharding, etc is not necessary, you could make a good case for the Kitchen Sink strategy.

The only change necessary to a standard Sitecore configuration to support this Kitchen Sink approach is to patch the contentSearch definitions for the Sitecore indexes where the name of the Solr “core” is specified (stored by default in config files like Sitecore.ContentSearch.Solr.Index.Master.config,  Sitecore.ContentSearch.Solr.Index.Web.config, etc).   This is telling Sitecore which Solr core contains the index, but the actual name of the core doesn’t factor into the ContentSearch API code one uses with Sitecore.   A patch such as the following would handle both the sitecore_master_index and the sitecore_web_index to organize into a Solr Core named “kitchen_sink:”

<configuration xmlns:patch="http://www.sitecore.net/xmlconfig/">
  <sitecore>
    <contentSearch>
      <configuration>
        <indexes>
          <index id="sitecore_master_index" type="Sitecore.ContentSearch.SolrProvider.SolrSearchIndex, Sitecore.ContentSearch.SolrProvider">
            <param desc="core">kitchen_sink</param>
          </index>
          <index id="sitecore_web_index" type="Sitecore.ContentSearch.SolrProvider.SolrSearchIndex, Sitecore.ContentSearch.SolrProvider">
            <param desc="core">kitchen_sink</param>
          </index>
        </indexes>
        </configuration>
    </contentSearch>
  </sitecore>
</configuration>

If you peek into the Solr Admin for the kitchen_sink core that I’m using, specifically the Schema Browser in the Solr Admin UI, it becomes clear how Sitecore uses a field named “_indexname” to represent the Sitecore index value.  For this screenshot below, I’ve set the kitchen_sink core to contain two Sitecore indexes: sitecore_master_index and sitecore_web index:

solrterms

This shows us the two terms stored in that _indexname field, and that there are 18,774 for sitecore_master_index and 5,851 for sitecore_web_index.  Even though the indexes are contained in the same Solr Core, Sitecore ContentSearch API code like this . . .

Sitecore.ContentSearch.ISearchIndex index = 
  ContentSearchManager.GetIndex(indexName);
    using (Sitecore.ContentSearch.IProviderSearchContext ctx = 
      index.CreateSearchContext())

. . . doesn’t care whether all the Sitecore indexes reside in a single Solr “Core” or if they’re in their own following a 1:1 mapping strategy.

Caveats and Going In A Different Direction

There was a bug or two in earlier versions of Sitecore related to this, so be careful with early Sitecore 7.2 or Sitecore 8 implementations (and if you’re using Sitecore 7.5, you’ve got plenty of other things to worry about so don’t sweat a Solr Core organization strategy!).

I should also note that while this post is looking at combining Sitecore indexes into a single Solr Core for convenience and to reduce the management headaches of having 15 sets of Solr Cores to update etc, there are some implementations that go in the opposite direction.  Consider a strategy like the following:

solrmindblown

 

There may be circumstances where keeping Sitecore indexes in their own Solr Core — and even isolating them further into their own Solr implementation — could be in order.  Solr runs in a JVM and this could certainly factor in, but there are other shared run-time resources that Solr sets aside for the whole Solr application.

I’m not familiar enough with these sorts of implementations that I want to comment further or recommend any course of action related to this right now, but it’s good to think about and consider with Solr tuning scenarios.  I just wanted to share it, as it’s a logical dimension to consider given the two previous strategies in this post.

 

The Solr *Optimize Now* Button for Sitecore Use Cases

If you’ve worked with Sitecore and Solr, you’re no stranger to the Solr Admin UI.  There are great aspects to that UI, and some exciting extension points with Sitecore implications too, but I know one element of that Solr UI that causes some head-scratching . . . the “optimize now” feature:

OptimizeNow.JPG

 

The inclusion of the badcauses people to think “something is wrong . . . red bad . . . must click now!”

What is this Optimize?

I’m writing this for the benefit of Sitecore developers who may not be coming at this from a deep search background: do not worry if your Solr cores show this badicon encouraging you to optimize now.  Fight that instinct.  For standard Sitecore Solr cores that are frequently updating, such as the sitecore_core_index, sitecore_master_index, sitecore_analytics_index, and — depending on your publishing strategy — sitecore_web_index, one may notice these cores almost always appear with this “optimize now” button in the Solr Admin UI.  Other indexes, too, may be heavily in use depending on how your Sitecore implementation is structured.  If you choose the optimize now option and then reload the screen, you’ll see the friendly green check mark next to Optimized and you should notice the Segment Count drops to a value of 1:

segmentcount1.JPG

Segments are Lucene’s (and therefore Solr’s) file system unit.  On disk, “segments” are where data is durably stored and organized for Lucene.  In Solr version 5 and newer, one can visualize Segment details for each Solr Core via the Solr Admin UI Segments Info screen.  This shows 2 Segments:

segmentinfo

If your Segment count is greater than 1, the Solr Admin UI will report that your Solr Core is in need of Optimization (with that somewhat alarmingbadicon).  The Optimize operation re-organizes all the Segments in a Core down to just one single Segment . . . and for busy Sitecore indexes this is not something to do very often (or at all!).

To track an optimize operation through at the file system level, consider this snapshot of the /data/index directory for a sitecore_master_index before performing optimization; note the quantity of files:optimizebefore

After the optimization, consider the same file system:

optimizeafter

When in doubt, don’t optimize

Solr’s optimize now command is like cleaning up a house after a party.  It reduces the clutter and consolidates the representation of the Solr Core on disk to a minimal footprint.  The problem, is, however, optimizing takes longer the larger the index is — so the act of optimizing may produce very non-optimal performance while it’s doing the work.  Solr has to read a copy of the entire index and restructure the copy into a single Segment.  This can be slow.  Caches must be re-populated after an optimization, too, compounding the perf impact.  To continue the analogy of the optimize now being like cleaning after a party, imagine cleaning up during a party; maybe you pause the music and ask everyone to leave the house for 20 minutes while you straighten everything up.  Then everyone returns and the partying resumes, with the cleaning being a mostly useless delay.

To draw from the official Solr documentation at https://wiki.apache.org/solr/SolrPerformanceFactors#Optimization_Considerations:

“Optimizing is very expensive, and if the index is constantly changing, the slight performance boost will not last long. The trade-off is not often worth it for a non static index.”

For those Sitecore indexes in Solr that are decidedly non-static, then, ignore that “optimize now” feature of the Solr Admin UI.  It’s better to pay attention to Solr “Merge Policies” for a rules based approach to maintaining Segments; this is a huge topic, one left for another time.

When to consider optimizing

Knowing more about the optimization process in Solr, then, we can think about when it may be appropriate to apply the optimize command.  For external data one is pulling into Solr, for example, a routine optimization could make sense.  If you have a weekly product data load, for instance, where 10,000 items are regularly loaded into a Solr Core and then they remain un-changed, optimization after the load completes makes a lot of sense.  That data in the Core is not dynamic.  When the data load completes, you could include an API call to Solr that triggers the optimize.

An API call to trigger an optimize in Solr is available through an update handler call : http://solr-server:8983/solr/sitecore_product_catalog/update?stream.body=<optimize><query>*:*</query></optimize&gt;

Sitecore search has a very checkered past with the Lucene Optimize operation.  I’ve worked on big projects that were crippled by too frequent optimizing work like that discussed in Uli Weltersbach’s post.  We ended up customizing the Optimize methods to be no-op statements, or another variation like that.  For additional validation, check out the Lucene docs on the optimize method:

“This method has been deprecated, as it is horribly inefficient and very rarely justified.”

Since Solr sits on top of Lucene, the heritage of Lucene’s optimize is still relevant and — in the Solr Admin UI — we see a potential performance bottleneck button ripe for clicking . . . fight that instinct!

now

 

Solr Configuration for Integration with Sitecore

I’ve got a few good Solr and Sitecore blogs around 75% finished, but I’ve been too busy lately to focus on finishing them.  In the meantime, I figure a picture can be worth 1,000 words sometimes so let me post this visual representation of Solr strategies for Sitecore integrations.  One Solr core per index is certainly the best practice for production Sitecore implementations, but now that Solr support has significantly matured at Sitecore a one Solr core for all the Sitecore indexes is a viable, if limited, option:

draft

There used to be a bug (or two?) that made this single Solr core for every Sitecore index unstable, but that’s been corrected for some time now.

More to follow!

Sitecore and Solr Diagnostics: Pings and Logs

Like I mentioned last week in this post about Sitecore + Solr – Lucene, I wanted to share some other posts about running Sitecore with Solr.  While 2017 is shaping up to be the year of Azure for Sitecore, I also expect Solr to continue to grow in prominence across Sitecore implementations.

Solr exposes lots of information about it’s current condition and state, but it’s not that easy to access.  I’ll take a run at shedding light on some of the monitoring basics and let’s see where this goes:

Ping

A proof-of-life for Solr is usually part of an implementation, whether it’s for a load-balancer in front of Solr nodes or some other form of a health check.  This is typically done via the Solr /admin/ping endpoint so a request to http://localhost:8983/solr/sitecore_master_index/admin/ping?wt=json would result in something like:

{
  "responseHeader": {
    "status": 0,
    "QTime": 0,
    "params": {
      "q": "{!lucene}*:*",
      "distrib": "false",
      "df": "text",
      "rows": "10",
      "wt": "json",
      "echoParams": "all"
    }
  },
  "status": "OK"
}

I chose the JSON response writer for Solr (more about this at https://wiki.apache.org/solr/CoreQueryParameters).  Solr’s default response writer is XML and returns a lot more formatting text around the exact same information.

There is actually a lot of flexibility around this ping RequestHandler for Solr (check out the documentation).  I’m scratching the surface here since, for a plain vanilla Sitecore install, knowing that ping will either be “OK” or return the failing HTTP error code is usually sufficient.

Logging

So the ping is easy, but only reports OK or NOT OK.  It doesn’t provide any real granularity around that determination . . . so let’s assume you’ve got a Solr server that is not returning an “OK” status from the ping.  If you’ve restarted Solr on the server, and Solr still isn’t responding “OK” then it’s probably time to dig into the Solr logs.  Solr defaults to using a variation of Log4j to capture trace logs, so the documentation on Log4j can be helpful (just note that currently Log4j is on version 2.x, while the Solr versions I’ve worked on [up to Solr 6.2] are all using the earlier Log4j 1.2 version [log4j-1.2.17.jar]).

Solr will use the file server\resources\log4j.properties as the basis for Log4j configuration, but this can vary depending on how you’ve installed Solr.   If you need to determine where this file is configured in your implementation, you can check the Dashboard of the Solr web admin tool.  This shows the runtime configuration settings for Solr.  I’ve highlighted the two log related settings in my example:

solrlogsettings

The -Xloggc parameter is for Java Garbage Collection (GC) logging and usually not something you would consider Solr specific, but good to know it exists if that becomes a topic to investigate.

Inside a standard log4j.properties file one finds the following:

#  Logging level
solr.log=logs
log4j.rootLogger=INFO, file

log4j.appender.CONSOLE=org.apache.log4j.ConsoleAppender

log4j.appender.CONSOLE.layout=org.apache.log4j.EnhancedPatternLayout
log4j.appender.CONSOLE.layout.ConversionPattern=%-4r %-5p (%t) [%X{collection} %X{shard} %X{replica} %X{core}] %c{1.} %m%n

#- size rotation with log cleanup.
log4j.appender.file=org.apache.log4j.RollingFileAppender
log4j.appender.file.MaxFileSize=50MB
log4j.appender.file.MaxBackupIndex=9

#- File to log to and log format
log4j.appender.file.File=${solr.log}/solr.log
log4j.appender.file.layout=org.apache.log4j.EnhancedPatternLayout
log4j.appender.file.layout.ConversionPattern=%d{yyyy-MM-dd HH:mm:ss.SSS} %-5p (%t) [%X{collection} %X{shard} %X{replica} %X{core}] %c{1.} %m\n

log4j.logger.org.apache.zookeeper=WARN
log4j.logger.org.apache.hadoop=WARN

# set to INFO to enable infostream log messages
log4j.logger.org.apache.solr.update.LoggingInfoStream=OFF

That configuration file is fairly self-explanatory.  Adjust the rootLogger value off of INFO if you’d prefer DEBUG or WARN or some other level.  I think I have just one other note on the above configuration file:

the MaxFileSize setting of older Solr installations was set to 4 MB; this is pretty small for a production instance, so the example above (using a Solr 6 default) of using 50 MB makes more sense for production.

If one peeks into a solr log file, there’s a couple common record types by default.  Here’s what a query logged for Solr looks like:

webapp=/solr path=/select params={df=text&distrib=false&fl=_uniqueid&fl=score&shards.purpose=4&start=0&fsv=true&fq=_indexname:(sitecore_testing_index)&shard.url=http://10.225.240.243:8983/solr/sitecore_testing_index_shard2_replica2/|http://10.225.240.72:8983/solr/sitecore_testing_index_shard2_replica1/&rows=2147483647&version=2&q=__is_running_b:(True)&NOW=1481640759343&isShard=true&wt=javabin} hits=0 status=0 QTime=0

After the query parameters, the log reports the number of hits for the query, the status (0 means “OK”) and the duration for this request in milliseconds.  Note that Solr queries are logged after they complete . . . so if you have a run-away query you may not see it in the log with the standard configuration.  More about this at a later time, perhaps.

Updates to Solr are logged using a similar pattern, they will start with webapp=/solr path=/update.

It’s possible to  setup Log4j to create more narrowly targeted logs.  For example, you could have queries and updates written to different logs for easier troubleshooting (see this example for the nitty gritty on targeting Solr logs).

If your project doesn’t want all this “noise” about each update and query to Solr, or there’s a concern that all this logging could become a bottleneck, you can set the log level for Solr to WARN and only see warnings and errors in your logs.  You will be sacrificing that query performance data, however, so you’ll not have a record to review for slow performing queries . . . so a recommended practice is to trigger a “WARN” condition when a query exceeds a certain threshold.  In solrconfig.xml add a <slowQueryThresholdMillis> element to the query section of solrconfig.xml:

<query>
<slowQueryThresholdMillis>2000</slowQueryThresholdMillis>

Any queries that take longer than 2 seconds (2000 milliseconds defined above) will be logged as “slow” queries at the WARN level.  Note that this setting would need to be duplicated for every core that’s defined in Solr (meaning, with Sitecore’s one core per index  approach you’ll have several solrconfig.xml files to touch).

Besides these fairly routine entries in Solr logs, there will also be exceptions and sometimes using the Logging section of the Solr web admin provides a convenient way to see the most recent set of these exceptions.  This is a convenient way of viewing Log4j items for your Solr implementation in near real-time; note if you make changes to the Level setting from this web interface, it is not persistent after a Solr restart.

Example of a Solr logging section in the web admin interface:

loggingsolradmin

There are a number of other ways to review the Log4j data, old-school Chainsaw is still around, some I know some prefer Otros log viewer, but I’ve had success with Splunk (see screen shot below):

splunkforSolr.JPG

I’ll get into other methods for diagnostic work with Solr in the next post.

Sitecore Gets Serious About Search, by Leaving the Game

Sitecore is steadily reducing the use-cases for using Lucene as the search provider for the Sitecore product.  Sitecore published this document on “Solr or Lucene” a few days ago, going so far as to state this one criteria where you must use Solr:

  • You have two or more content delivery servers

Read that bullet point again.  If you’ve worked with Sitecore for a while, this should trigger an alarm.

Sound the alarm: most implementations have 2 or more CD servers!  I’d say 80% or more of Sitecore implementations are using more than a single CD server, in fact.  Carrying this logic forward, 80% of Sitecore implementations should not be running on Lucene!  This is a big departure for Sitecore as a company, who would historically dodge conversations about what is the right technology for a given situation.  I think the Sitecore philosophy was that advocating for one technical approach over another is risky and means a Sitecore endorsement could make them accountable if a technology approach fails for one reason or another.  For as long as I’ve known Sitecore, it’s a software company intent on selling licenses, not dictating how to use the product.  Risk aversion has kept them from getting really down in the weeds with customers.  This tide has been turning, however, with things like Helix and now this more aggressive messaging about the limitations of Lucene.  I think it’s great for Sitecore to be more vocal about best practices, it’s just taken years for them to come around to the idea.

As a bit of a search geek, I want to state for the record that this new Solr over Lucene guidance from Sitecore is not really an indictment of Lucene.  The Apache Lucene project, and it’s cousin the .Net port Lucene.net that Sitecore makes use of out-of-the-box, was ground breaking in many ways.  As a technology, Lucene can handle enormous volumes of documents and performs really well.  Solr is making use of Lucene underneath it all, anyway!  This recent announcement from Sitecore is more acknowledgement that Sitecore’s event plumbing is no substitute for Solr’s CAP-theorem straddling acrobatics.  Sitecore is done trying to roll their own distributed search system.  I think this is Sitecore announcing that they’re tired of patching the EventQueue, troubleshooting search index update strategies for enterprise installations, and giving up on ensuring clients don’t hijack the Sitecore heartbeat thread and block search indexing with some custom boondoggle in the publishing pipeline.  They’re saying: we give up — just use Solr.

Amen.  I’m fine with that.

To be honest, I think this change of heart can also be explained by the predominant role Azure Search plays in the newest Sitecore PaaS offering.  Having an IP address for all the “search stuff” is awful nice whether you’re running in Azure or anywhere else.  It’s clear Sitecore decided they weren’t keen to recreate the search wheel a few years ago, and are steadily converging around these technologies with either huge corporate backing (Azure Search) or a vibrant open source community (Solr).

I should also note, here, that Coveo for Sitecore probably welcomes this opportunity for Sitecore teams to break away from the shackles of the local Lucene index.  I’m not convinced that long-term Coveo can out-run the likes of Azure Search or Solr, but I know today if your focus is quickly ramping up a search heavy site on Sitecore you should certainly weight the pros/cons of Coveo.

With all this said, I want to take my next few posts and dig really deep into Solr for Sitecore and talk about performance monitoring, tuning, and lots of areas that get overlooked when it comes to search with Sitecore and Solr.   So I’ll have more to say on this topic in the next few days!

 

Slice & Dice Solr Indexes for Sitecore with Banana UI

I’m fresh off a week at the Lucene-Solr Revolution conference and training, and it was a real eye-opener for me!  I have many pages of notes, so many potential blog posts bridging the Solr/Sitecore divide, and saw up-close the vibrancy of the Solr community.  There’s so much innovation going on there . . . it was really inspiring.

One take away from the experience was how Solr has a variety of tools for index analysis that could be really useful for Sitecore scenarios.  I know there’s some basic Sitecore diagnostic pages that will shed light on some index internals for Lucene or Solr, and the Solr admin UI has lots of powerful features, but I wanted to share in this blog post a bit about Banana UI and what it can do for us in the Sitecore world.

What Banana UI Can Do

Before explaining how Banana UI works, it’s probably most useful to show a screen shot of it in action:

banana0

The above shows 3 custom dashboard widgets (Banana calls them panels) that are processing against my sitecore_master_index for a LaunchSitecore 8.1 demo instance I have.  I quickly contrived the above 3 visualizations, using just some sample data, but they represent a bit of the possible options using this free tool.

The first dashboard charts parsedcreatedby_s values for the LaunchSitecore site (I set it to exclude the 15,000 items authored by the Sitecore admin using a Solr filter of q=-parsedupdatedby_s:sitecoreadmin).

banana2.JPG

The second dashboard is a basic word cloud derived from title_t in the LaunchSitecore site index.  Banana UI makes it easy to inspect the underlying query for a visualization, so I could find the baseline query executed against Solr for this was q=*%3A*&wt=json&rows=0&fq=__smallcreateddate_tdt:[2011-10-17T19:49:57.000Z%20TO%202016-10-17T19:49:57.000Z]&facet=true&facet.field=title_t&facet.limit=20).

banana3.JPG

The third dashboard shows a heat map of culture_s and menu_title_t.  I wanted to show the heatmap since it’s a great way of visualizing data correlations, and in this case it illustrates how heavily LaunchSitecore relies on the en culture for item content.

banana4

The crux of the Banana UI is time series of data, although it has powerful applications for other data too.  For my example dashboards, I setup a time window spanning 5 years so it included all the origin data for LaunchSitecore (has it really been 5 years since that project got off the ground?).  Like I said above, I wanted to exclude the 15,000+ pieces of sitecore\admin content since it would dominate the chart in dashboard 1.  With Banana UI, it’s all driven through the single page application:

banana1.JPG

I’m just getting started with Banana as a tool for examining customer Solr implementations for Sitecore . . . I can envision a lot of interesting angles to this, whether using the sitecore_analytics_index for xDB visualizations or metadata analysis with canned searches of the sitecore_web_index.  There’s probably a use case for everyone: business users wanting to keep tabs on a few key terms in the corpus, super-users doing site-wide analysis, certainly for technical users curious about term frequencies or looking to test out some Solr faceting with zero coding and a nice visual interface for feedback.  Banana UI belongs on the radar of every Sitecore implementation making use of Solr.

How To Put Banana UI To Use

Quoting from the Lucidwork’s GitHub repo for Banana:

The Banana project was forked from Kibana, and works with all kinds of time series (and non-time series) data stored in Apache Solr. It uses Kibana’s powerful dashboard configuration capabilities, ports key panels to work with Solr, and provides significant additional capabilities, including new panels that leverage D3.js.

The goal is to create a rich and flexible UI, enabling users to rapidly develop end-to-end applications that leverage the power of Apache Solr.

Here’s a key reference on how to generally configure Banana for Solr.

I’m not going to repeat the reference above, but the gist of it is to copy the Banana UI source into your solr-webapp directory and customize a couple settings (like the name of the Solr collection you want to analyze), then you’re ready to begin slicing and dicing the Solr information:

banana

One can access the Banana UI  via http://your solr server:your solr port/solr/banana/ (or solr/banana/src/index.html if the redirects aren’t setup).  I found Chrome to be the best browser for visualizing the data.  You can save pre-built dashboards for ease of sharing, or build dashboards out ad-hoc.

While Banana (and the original, Kibana) is designed for viewing event and log data in real-time, it’s charting components work great as a facade in front of Solr and it deploys as a simple stand-alone directory as part of the Solr web app.  I’m looking forward to seeing how I can further shape Sitecore specific dashboards with Banana, as currently I’m really just scratching the surface.

The Sitecore Pie: strategic slicing for better implementations

The Sitecore Pie

pie
At some point I want to catalog the full set of features a Sitecore Content Delivery and Content Management server can manage in a project.  My goal would be to identify all the elements that can be split apart into independent services.  This post is not a comprehensive list of those features, but serves to introduce the concept.

Think of Sitecore as a big blueberry pie that can be sliced into constituent parts.  Some Sitecore sites can really benefit from slicing the pie into small pieces and letting dedicated servers or services manage that specific aspect of the pie.  Too often, companies don’t strategize around how much different work their Sitecore solution is doing.

An example will help communicate my point: consider IIS and how it serves as the execution context for Sitecore.   Many implementations will run logic for the following through the same IIS server that is handling the Sitecore request for rendering a web page.  These are all slices of the Sitecore pie for a Sitecore Content Delivery server:

  1. URL redirection through Sitecore
  2. Securing HTTP traffic with SSL
  3. Image resizing for requests using low-bandwidth or alternative devices
  4. Serving static assets like CSS, JS, graphics, etc
  5. Search indexing and query processing (if one is using Lucene)

If you wanted to cast a broader net, you could include HTTP Session state for when InProc mode is chosen, Geo-IP look-ups for certain CD servers, and others to this list of pie slices.  Remember, I didn’t claim this was an exhaustive list.  The point is: IIS is enlisted in all this other work besides processing content into HTML output for Sitecore website visitors.

Given our specific pie slices above, one could employ the following alternatives to relieve IIS of the processing:

  1. URL Redirection at the load balancer level can be more performant than having Sitecore process redirects
  2. Apply SSL between the load balancer and the public internet, but not between the IIS nodes behind your load balancer — caled “SSL Offloading” or “SSL Termination”
  3. There are services like Akamai that fold in dynamic image processing as part of their suite of products
  4. Serving static assets from a CDN is common practice for Sitecore
  5. Coveo for Sitecore is an alternative search provider that can take a lot of customer-facing search aspects and shift it to dedicated search servers or even Coveo’s Cloud.  One can go even further with Solr for Sitecore or still other search tiers if you’re really adventurous

My point is, just like how we hear a lot this election season about “let Candidate X be Candidate X” — we can let Sitecore be Sitecore and allow it to focus on rendering content created and edited by content authors and presenting it as HTML responses.  That’s what Sitecore is extremely valuable for.

Enter the Cache

I’ve engaged with a lot of Sitecore implementations who were examining their Sitecore pie and determining what slices belong where . . . and frequently we’d make the observation that the caching layer of Sitecore was tightly coupled with the rest of the Sitecore system and caching wasn’t a good candidate for slicing off.  There wasn’t a slick provider model for Sitecore caches, and while certain caches could be partially moved to other servers, it wasn’t clean, complete, or convenient.

That all changed officially with the initial release of Sitecore 8.2 last month.  Now there is a Sitecore.Caching.DefaultCacheManager class, a Sitecore.Caching.ICache interface, and other key extension points as part of the standard Sitecore API.  One can genuinely add the Sitecore cache to the list of pie slices one can consider for off-loading.

In my next post, I will explore using this new API to use Redis as the cache provider for Sitecore instead of the standard in-memory IIS cache.

Search User Group Talk

I did a talk for the New England Search Technology (NEST) user group yesterday.  Even though the meetings in Boston are a good 90+ minutes away for me, I try to make the trip there a couple times a year since the topics are usually very relevant to what I’m up to with Sitecore.   I offered to do a talk bridging the Sitecore and Search domains, and they took me up on it.  The audience is typically serious Solr and ElasticSearch technologists, some regular committers to those projects, so it was fun to combine those domains with Sitecore’s relative immaturity when it comes to the platform of search.

I don’t want to just post the powerpoint presentations and say “have at it” since the presentations require context to make sense (and it is 2 different powerpoints to sift through).  I’m a non-conventional guy, and I try to avoid powerpoint with bullet points galore and hyper-structured material.  The talk was more a conversation about search and the challenges unique to search with Sitecore (and other CMS platforms that build on Lucene).

My premise relied heavily on Plato’s Allegory of the Cave where I was a “prisoner” experiencing the world of search through the “cave” of Sitecore.  In reality, search is an enormous space with lots of complexity and innovation across the technology . . . but in terms of Sitecore, we experience a filtered reality (the shadows on the cave wall).  This graphic represents a traditional Allegory of the Cave concept:

cave

I’m not going to summarize the entire talk, but I want to state for the record this isn’t a particular criticism of Sitecore — it’s just the nature of working with a product that is built on top of other products.  In Sitecore’s case with Lucene, for example, there’s a .Net port of Lucene (Lucene.Net) and a custom .Net event pipeline in Sitecore used to orchestrate Lucene.Net activities via an execution context of IIS, and so on.

Understanding Search technology with Sitecore is always going to be filtered through the lens of Sitecore.  My talk was addressing that fact, and how Search (with a capital “S”) is a far broader technology than what is put to use in Sitecore.  Understanding the world of Search beyond the confines of Sitecore can be very revealing, and opens up a lot of opportunity for improving a Sitecore implementation.

I should also note that my talk benefited from insightful input from Sitecore-Coveo Jeff, Tim “Sitecore MVP” Braga, and Al Cole from Lucidworks and others.  The host, Monster.com, shared a great meeting space for us and opened the evening with a quick survey of their specific experience with search using both Solr, Elasticsearch, and their own proprietary technology.

While not specific to Sitecore, one link I wanted to share in particular was the talk about Bloomberg’s 3-year journey with Solr/Lucene; it’s a talk from the Berlin Buzzwords conference a couple weeks ago and thoroughly worth watching.  Getting search right can take persistence and smart analysis, regardless of the platform.  With Sitecore, too often implementations assume Search will just work out of the box and not appreciate that it’s a critical set of features worthy of careful consideration.

I’ll have a few follow-up posts covering more of the points I made in my talk; some lend themselves to distinct blog posts instead of turning this into a sprawling re-hash of the entire evening.