You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@accumulo.apache.org by el...@apache.org on 2017/03/06 23:00:51 UTC
[2/2] accumulo-website git commit: Jekyll build from master:5679ae1
Jekyll build from master:5679ae1
Wrote blog post about security and performance.
Closes apache/accumulo-website#8
Signed-off-by: Josh Elser <el...@apache.org>
Project: http://git-wip-us.apache.org/repos/asf/accumulo-website/repo
Commit: http://git-wip-us.apache.org/repos/asf/accumulo-website/commit/803c95d0
Tree: http://git-wip-us.apache.org/repos/asf/accumulo-website/tree/803c95d0
Diff: http://git-wip-us.apache.org/repos/asf/accumulo-website/diff/803c95d0
Branch: refs/heads/asf-site
Commit: 803c95d0cd3065eead750ad2027bc49f57c11c87
Parents: f183655
Author: Josh Elser <el...@apache.org>
Authored: Mon Mar 6 17:58:45 2017 -0500
Committer: Josh Elser <el...@apache.org>
Committed: Mon Mar 6 17:58:45 2017 -0500
----------------------------------------------------------------------
README.md | 1 -
.../23/security-performance-implications.html | 449 +++++++++++++++++++
feed.xml | 385 ++++++++++++----
images/blog/201702_security/accumuloFiles.png | Bin 0 -> 313628 bytes
images/blog/201702_security/avgFilesTab.png | Bin 0 -> 288346 bytes
images/blog/201702_security/figure1.png | Bin 0 -> 78725 bytes
images/blog/201702_security/tableRecs.png | Bin 0 -> 159556 bytes
images/blog/201702_security/tableRecsInMem.png | Bin 0 -> 223097 bytes
images/blog/201702_security/totalIngest.png | Bin 0 -> 245159 bytes
index.html | 14 +-
news/index.html | 7 +
11 files changed, 754 insertions(+), 102 deletions(-)
----------------------------------------------------------------------
http://git-wip-us.apache.org/repos/asf/accumulo-website/blob/803c95d0/README.md
----------------------------------------------------------------------
diff --git a/README.md b/README.md
index 9d4b0ef..13e7cc0 100644
--- a/README.md
+++ b/README.md
@@ -28,4 +28,3 @@ This can be done easily by invoking the post-commit hook (either by hand, or aut
Git to invoke the post-commit hook).
`./_devtools/git-hooks/post-commit`
-
http://git-wip-us.apache.org/repos/asf/accumulo-website/blob/803c95d0/blog/2017/02/23/security-performance-implications.html
----------------------------------------------------------------------
diff --git a/blog/2017/02/23/security-performance-implications.html b/blog/2017/02/23/security-performance-implications.html
new file mode 100644
index 0000000..fd23824
--- /dev/null
+++ b/blog/2017/02/23/security-performance-implications.html
@@ -0,0 +1,449 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+<!--
+ Licensed to the Apache Software Foundation (ASF) under one or more
+ contributor license agreements. See the NOTICE file distributed with
+ this work for additional information regarding copyright ownership.
+ The ASF licenses this file to You under the Apache License, Version 2.0
+ (the "License"); you may not use this file except in compliance with
+ the License. You may obtain a copy of the License at
+
+ http://www.apache.org/licenses/LICENSE-2.0
+
+ Unless required by applicable law or agreed to in writing, software
+ distributed under the License is distributed on an "AS IS" BASIS,
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ See the License for the specific language governing permissions and
+ limitations under the License.
+-->
+<meta charset="utf-8">
+<meta http-equiv="X-UA-Compatible" content="IE=edge">
+<meta name="viewport" content="width=device-width, initial-scale=1">
+<link href="https://maxcdn.bootstrapcdn.com/bootswatch/3.3.7/paper/bootstrap.min.css" rel="stylesheet" integrity="sha384-awusxf8AUojygHf2+joICySzB780jVvQaVCAt1clU3QsyAitLGul28Qxb2r1e5g+" crossorigin="anonymous">
+<link href="//netdna.bootstrapcdn.com/font-awesome/4.0.3/css/font-awesome.css" rel="stylesheet">
+<link rel="stylesheet" type="text/css" href="https://cdn.datatables.net/v/bs/jq-2.2.3/dt-1.10.12/datatables.min.css">
+<link href="/css/accumulo.css" rel="stylesheet" type="text/css">
+
+<title>Security Performance Implications</title>
+
+<script src="https://ajax.googleapis.com/ajax/libs/jquery/2.2.4/jquery.min.js"></script>
+<script src="https://maxcdn.bootstrapcdn.com/bootstrap/3.3.7/js/bootstrap.min.js" integrity="sha384-Tc5IQib027qvyjSMfHjOMaLkfuWVxZxUPnCJA7l2mCWNIpG9mGCD8wGNIcPD7Txa" crossorigin="anonymous"></script>
+<script type="text/javascript" src="https://cdn.datatables.net/v/bs/jq-2.2.3/dt-1.10.12/datatables.min.js"></script>
+<script>
+ // show location of canonical site if not currently on the canonical site
+ $(function() {
+ var host = window.location.host;
+ if (typeof host !== 'undefined' && host !== 'accumulo.apache.org') {
+ $('#non-canonical').show();
+ }
+ });
+
+ $(function() {
+ // decorate section headers with anchors
+ return $("h2, h3, h4, h5, h6").each(function(i, el) {
+ var $el, icon, id;
+ $el = $(el);
+ id = $el.attr('id');
+ icon = '<i class="fa fa-link"></i>';
+ if (id) {
+ return $el.append($("<a />").addClass("header-link").attr("href", "#" + id).html(icon));
+ }
+ });
+ });
+
+ // configure Google Analytics
+ (function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){
+ (i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o),
+ m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m)
+ })(window,document,'script','//www.google-analytics.com/analytics.js','ga');
+
+ if (ga.hasOwnProperty('loaded') && ga.loaded === true) {
+ ga('create', 'UA-50934829-1', 'apache.org');
+ ga('send', 'pageview');
+ }
+</script>
+
+</head>
+<body style="padding-top: 100px">
+
+ <nav class="navbar navbar-default navbar-fixed-top">
+ <div class="container">
+ <div class="navbar-header">
+ <button type="button" class="navbar-toggle" data-toggle="collapse" data-target="#navbar-items">
+ <span class="sr-only">Toggle navigation</span>
+ <span class="icon-bar"></span>
+ <span class="icon-bar"></span>
+ <span class="icon-bar"></span>
+ </button>
+ <a href="/"><img id="nav-logo" alt="Apache Accumulo" class="img-responsive" src="/images/accumulo-logo.png" width="200"/></a>
+ </div>
+ <div class="collapse navbar-collapse" id="navbar-items">
+ <ul class="nav navbar-nav">
+ <li class="nav-link"><a href="/downloads">Download</a></li>
+ <li class="dropdown">
+ <a class="dropdown-toggle" data-toggle="dropdown" href="#">Releases<span class="caret"></span></a>
+ <ul class="dropdown-menu">
+ <li><a href="/release/accumulo-1.8.1/">1.8.1 (Latest)</a></li>
+ <li><a href="/release/accumulo-1.7.2/">1.7.2</a></li>
+ <li><a href="/release/accumulo-1.6.6/">1.6.6</a></li>
+ <li><a href="/release/">Archive</a></li>
+ </ul>
+ </li>
+ <li class="dropdown">
+ <a class="dropdown-toggle" data-toggle="dropdown" href="#">Documentation<span class="caret"></span></a>
+ <ul class="dropdown-menu">
+ <li><a href="/1.8/accumulo_user_manual.html">User Manual (1.8)</a></li>
+ <li><a href="/1.8/apidocs">Javadocs (1.8)</a></li>
+ <li><a href="/1.8/examples">Examples (1.8)</a></li>
+ <li><a href="/features">Features</a></li>
+ <li><a href="/glossary">Glossary</a></li>
+ <li><a href="/external-docs">External Docs</a></li>
+ <li><a href="/docs-archive/">Archive</a></li>
+ </ul>
+ </li>
+ <li class="dropdown">
+ <a class="dropdown-toggle" data-toggle="dropdown" href="#">Community<span class="caret"></span></a>
+ <ul class="dropdown-menu">
+ <li><a href="/get_involved">Get Involved</a></li>
+ <li><a href="/mailing_list">Mailing Lists</a></li>
+ <li><a href="/people">People</a></li>
+ <li><a href="/related-projects">Related Projects</a></li>
+ <li><a href="/contributor/">Contributor Guide</a></li>
+ </ul>
+ </li>
+ </ul>
+ <ul class="nav navbar-nav navbar-right">
+ <li class="dropdown">
+ <a class="dropdown-toggle" data-toggle="dropdown" href="#">Apache Software Foundation<span class="caret"></span></a>
+ <ul class="dropdown-menu">
+ <li><a href="https://www.apache.org">Apache Homepage <i class="fa fa-external-link"></i></a></li>
+ <li><a href="https://www.apache.org/licenses/LICENSE-2.0">License <i class="fa fa-external-link"></i></a></li>
+ <li><a href="https://www.apache.org/foundation/sponsorship">Sponsorship <i class="fa fa-external-link"></i></a></li>
+ <li><a href="https://www.apache.org/security">Security <i class="fa fa-external-link"></i></a></li>
+ <li><a href="https://www.apache.org/foundation/thanks">Thanks <i class="fa fa-external-link"></i></a></li>
+ <li><a href="https://www.apache.org/foundation/policies/conduct">Code of Conduct <i class="fa fa-external-link"></i></a></li>
+ </ul>
+ </li>
+ </ul>
+ </div>
+ </div>
+</nav>
+
+
+ <div class="container">
+ <div class="row">
+ <div class="col-md-12">
+
+ <div id="non-canonical" style="display: none; background-color: #F0E68C; padding-left: 1em;">
+ Visit the official site at: <a href="https://accumulo.apache.org">https://accumulo.apache.org</a>
+ </div>
+ <div id="content">
+
+ <h1 class="title">Security Performance Implications</h1>
+
+ <table>
+
+<tr><td><b>Date </b></td><td> 23 Feb 2017 </td></tr>
+
+</table>
+<p>
+
+
+<p>The purpose of this two part series was to measure the performance impact of
+various security configurations on a cluster running Apache Accumulo\u2019s
+continuous ingest suite. The tests were performed using Amazon Web
+Services (AWS), Hortonworks Data Platform 2.4 and Accumulo 1.7. Each of
+the five different security settings in Accumulo 1.7 was tested including
+no security, SSL, and SASL with Kerberos authentication for the three quality
+of protection levels (auth, auth-int, auth-conf). KDC was MIT. HDFS was
+configured to use Kerberos for authentication and had service level
+authorization on. Other than that, no other security settings (HTTPS, RPC
+protection, data transfer encryption, etc) were enabled. Timely was a
+separate, single node HDFS/Zookeeper/Accumulo instance.</p>
+
+<h2 id="intro">Intro</h2>
+
+<p>All runs utilized the continuous ingest suite that ships with Accumulo (a
+standard method to measure performance in Accumulo). It generates random
+graph data and inserts it into Accumulo, creating
+a long linked list of entries. Part 1 was run with just continuous ingest.<br />
+Based on the test results, there was a measurable performance impact as each additional security configuration was put in place.</p>
+
+<h2 id="methodology">Methodology</h2>
+
+<p>We ran 5 tests, one for each security configuration. Each iteration of each test inserted 2 billion entries. Batch writers were configured with 500K max mem
+to artificially inflate the overall write overhead. This was performed on a
+small cluster on AWS.</p>
+
+<p>Each test used one of the following security configurations:</p>
+
+<ul>
+ <li>No security - Default</li>
+ <li>Two way SSL</li>
+ <li>Kerberos/SASL with auth
+ <ul>
+ <li>auth is just Kerberos authentication between client and server. Each end of the RPC definitively knows who the other is.</li>
+ </ul>
+ </li>
+ <li>Kerberos/SASL with auth-int
+ <ul>
+ <li>Builds on auth, also providing message integrity checks of the data going across the wire. You also know that the message you received was not altered.</li>
+ </ul>
+ </li>
+ <li>Kerberos/SASL with auth-conf
+ <ul>
+ <li>Builds on auth-int, also providing confidentiality of the message that was sent to prevent others from reading it (aka wire-encryption).</li>
+ </ul>
+ </li>
+</ul>
+
+<p>For each test, five iterations were run to obtain a min, max, and median
+time elapsed at each security configuration. After each iteration,
+Hadoop, and Zookeeper processes were restarted, Accumulo tables are
+wiped clean and tables are recreated. In addition, pagecache, dentries
+and inodes are dropped by issuing a \u20183\u2019 command on
+/proc/sys/vm/drop_caches to ensure that the OS is not caching things to disk
+that might affect the benchmark. The following sequence was performed
+between iterations:</p>
+
+<ol>
+ <li>Bring down Accumulo</li>
+ <li>Bring down Zookeeper</li>
+ <li>Bring down Hadoop</li>
+ <li>Run sync command</li>
+ <li>Drop OS cache</li>
+ <li>Bring up Hadoop</li>
+ <li>Bring up Zookeeper</li>
+ <li>Bring up Accumulo</li>
+ <li>Drop tables</li>
+ <li>Create tables</li>
+</ol>
+
+<p>For each iteration, the results were stored, fed into <a href="https://nationalsecurityagency.github.io/timely/">Timely</a>, and viewed with Grafana.
+Since the runs were executed sequentially, the start epochs for each run did not align.
+To mitigate, the entries for each run were inserted
+with the same relative epoch for convenient comparison in Grafana.</p>
+
+<p>The table configurations for Accumulo remain the same throughout the
+different iterations and security levels. The Accumulo site
+configurations differ only due to the different settings for the
+security level configurations.</p>
+
+<h2 id="environment">Environment</h2>
+
+<p>In order to perform the testing, a small AWS cluster was setup using 14
+hosts on EC2. Two i2.xlarge instances were used as master nodes and eight
+d2.xlarge instances were used for workers. In addition, two c4.4xlarge
+instances were used for ingesters, one m4.2xlarge instance was used for
+Timely, and one m4.xlarge instance was used for Apache Ambari. A logical
+diagram of the setup is depicted below:</p>
+
+<p><img src="/images/blog/201702_security/figure1.png" alt="" width="400px" /></p>
+
+<p>Figure 1 - Cluster Layout, Roles, and Instance Types on AWS.</p>
+
+<p>The types of nodes and their function are given below:</p>
+
+<table id="instance_types" class="table">
+ <thead>
+ <tr>
+ <th style="text-align: left">Node Type</th>
+ <th style="text-align: left">AWS EC2 Type</th>
+ <th style="text-align: left">EC2 Type Details</th>
+ <th style="text-align: left">Quantity</th>
+ </tr>
+ </thead>
+ <tbody>
+ <tr>
+ <td style="text-align: left">Ingest Nodes</td>
+ <td style="text-align: left">c4.4xlarge</td>
+ <td style="text-align: left">16 core, 30 GB RAM</td>
+ <td style="text-align: left">2</td>
+ </tr>
+ <tr>
+ <td style="text-align: left">Worker Node</td>
+ <td style="text-align: left">d2.xlarge</td>
+ <td style="text-align: left">4 cores, 30.5 GB RAM, 3x2T GB HD</td>
+ <td style="text-align: left">8</td>
+ </tr>
+ <tr>
+ <td style="text-align: left">Master Node</td>
+ <td style="text-align: left">i2.xlarge</td>
+ <td style="text-align: left">4 cores, 30.5 GB RAM, 1x800GB SSD</td>
+ <td style="text-align: left">2</td>
+ </tr>
+ <tr>
+ <td style="text-align: left">Admin Node</td>
+ <td style="text-align: left">m4.xlarge</td>
+ <td style="text-align: left">4 cores, 16 GB RAM</td>
+ <td style="text-align: left">1</td>
+ </tr>
+ <tr>
+ <td style="text-align: left">Timely Node</td>
+ <td style="text-align: left">m4.2xlarge</td>
+ <td style="text-align: left">8 cores, 32 GB RAM</td>
+ <td style="text-align: left">1</td>
+ </tr>
+ </tbody>
+</table>
+
+<p>Table 1 \u2013 AWS Instance Types, Role, Details, and Quantities</p>
+
+<h2 id="results">Results</h2>
+
+<p>The median, max, and min of the milliseconds elapsed
+time of all iterations for each test is displayed below. The percentage change
+columns compare the Median, Max, and Min respectively from the no
+security level to each security configuration (e.g. no security Median
+vs. auth-int Median, no security Max vs. auth-int Max).</p>
+
+<table id="results" class="table">
+ <thead>
+ <tr>
+ <th>Security Level</th>
+ <th style="text-align: right">Median</th>
+ <th style="text-align: right">Standard Deviation</th>
+ <th style="text-align: right">Max</th>
+ <th style="text-align: right">Min</th>
+ <th style="text-align: right">% Change (nosec Median vs. Median)</th>
+ <th style="text-align: right">% Change (nosec Max vs. Max)</th>
+ <th style="text-align: right">% Change (nosec Min vs. Min)</th>
+ <th style="text-align: right">Delta from Previous Level (Median)</th>
+ </tr>
+ </thead>
+ <tbody>
+ <tr>
+ <td>no security</td>
+ <td style="text-align: right">7829394</td>
+ <td style="text-align: right">139340</td>
+ <td style="text-align: right">8143035</td>
+ <td style="text-align: right">7764309</td>
+ <td style="text-align: right">0.00%</td>
+ <td style="text-align: right">0.00%</td>
+ <td style="text-align: right">0.00%</td>
+ <td style="text-align: right">0.00%</td>
+ </tr>
+ <tr>
+ <td>ssl</td>
+ <td style="text-align: right">8292760</td>
+ <td style="text-align: right">87012</td>
+ <td style="text-align: right">8464060</td>
+ <td style="text-align: right">8204955</td>
+ <td style="text-align: right">5.92%</td>
+ <td style="text-align: right">3.94%</td>
+ <td style="text-align: right">5.68%</td>
+ <td style="text-align: right">5.92%</td>
+ </tr>
+ <tr>
+ <td>auth</td>
+ <td style="text-align: right">8859552</td>
+ <td style="text-align: right">134109</td>
+ <td style="text-align: right">9047971</td>
+ <td style="text-align: right">8657618</td>
+ <td style="text-align: right">13.16%</td>
+ <td style="text-align: right">11.11%</td>
+ <td style="text-align: right">11.51%</td>
+ <td style="text-align: right">6.83%</td>
+ </tr>
+ <tr>
+ <td>auth-int</td>
+ <td style="text-align: right">9500737</td>
+ <td style="text-align: right">155968</td>
+ <td style="text-align: right">9753424</td>
+ <td style="text-align: right">9282371</td>
+ <td style="text-align: right">21.34%</td>
+ <td style="text-align: right">19.78%</td>
+ <td style="text-align: right">19.55%</td>
+ <td style="text-align: right">7.24%</td>
+ </tr>
+ <tr>
+ <td>auth-conf</td>
+ <td style="text-align: right">9479635</td>
+ <td style="text-align: right">170823</td>
+ <td style="text-align: right">9776580</td>
+ <td style="text-align: right">9282189</td>
+ <td style="text-align: right">21.08%</td>
+ <td style="text-align: right">20.06%</td>
+ <td style="text-align: right">19.55%</td>
+ <td style="text-align: right">-0.22%</td>
+ </tr>
+ </tbody>
+</table>
+
+<p>Table 2 \u2013 Summarized Time Elapsed for Each Security Level</p>
+
+<h2 id="plots">Plots</h2>
+
+<p>Below are some snapshots of *stats.out elements via Grafana that were inserted
+into Timely with the same relative start time. Each graph represents a field
+in the output generated by <a href="https://github.com/apache/accumulo/blob/1.7/test/src/main/java/org/apache/accumulo/test/continuous/ContinuousStatsCollector.java">ContinuousStatsCollector</a></p>
+
+<h3 id="tablerecshttpsgithubcomapacheaccumuloblob17coresrcmainjavaorgapacheaccumulocoremasterthrifttableinfojaval73"><a href="https://github.com/apache/accumulo/blob/1.7/core/src/main/java/org/apache/accumulo/core/master/thrift/TableInfo.java#L73">TABLE_RECS</a></h3>
+<p>(Number of records in the continuous ingest table. Down sample=1m, aggregate=avg)</p>
+
+<p><a href="/images/blog/201702_security/tableRecs.png"><img src="/images/blog/201702_security/tableRecs.png" alt="" width="800px" /></a></p>
+
+<h3 id="totalingesthttpsgithubcomapacheaccumuloblob17coresrcmainjavaorgapacheaccumulocoremasterthrifttableinfojaval77"><a href="https://github.com/apache/accumulo/blob/1.7/core/src/main/java/org/apache/accumulo/core/master/thrift/TableInfo.java#L77">TOTAL_INGEST</a></h3>
+<p>(Ingest rate for Accumulo instance. Down sample=5m, aggregate=avg)</p>
+
+<p><a href="/images/blog/201702_security/totalIngest.png"><img src="/images/blog/201702_security/totalIngest.png" alt="" width="800px" /></a></p>
+
+<h3 id="avgfilestablethttpsgithubcomapacheaccumuloblob17coresrcmainjavaorgapacheaccumulocoreutilstatjaval63"><a href="https://github.com/apache/accumulo/blob/1.7/core/src/main/java/org/apache/accumulo/core/util/Stat.java#L63">AVG_FILES/TABLET</a></h3>
+<p>(Average number of files per Accumulo tablet. Down sample=1m, aggregate=avg)</p>
+
+<p><a href="/images/blog/201702_security/avgFilesTab.png"><img src="/images/blog/201702_security/avgFilesTab.png" alt="" width="800px" /></a></p>
+
+<h3 id="accumulofileshttpsgithubcomapacheaccumuloblob17testsrcmainjavaorgapacheaccumulotestcontinuouscontinuousstatscollectorjaval127"><a href="https://github.com/apache/accumulo/blob/1.7/test/src/main/java/org/apache/accumulo/test/continuous/ContinuousStatsCollector.java#L127">ACCUMULO_FILES</a></h3>
+<p>(Total number of files for Accumulo. Down sample=1m, aggregate=avg)</p>
+
+<p><a href="/images/blog/201702_security/accumuloFiles.png"><img src="/images/blog/201702_security/accumuloFiles.png" alt="" width="800px" /></a></p>
+
+<p>As can be seen in the plots above, the different security settings have
+relatively consistent, discernable median run characteristics. The big
+dip in each TOTAL_INGEST coincides with a large number of major
+compactions, a rate decrease for TABLE_RECS, and a decrease in
+AVG_FILES/TABLET.</p>
+
+<h2 id="final-thoughts">Final Thoughts</h2>
+
+<p>The biggest performance
+hits to run duration median (compared to default security) were ~21% for
+auth-int and auth-conf. Interesting to note that SSL\u2019s median run duration was
+lower than all SASL configs and that auth-conf\u2019s was lower than auth-int.
+Initial speculation for these oddities revolved around the
+<a href="https://github.com/m1ch1/mapkeeper/wiki/Thrift-Java-Servers-Compared">Thrift server</a>
+implementations, but the Thrift differences will not explain the auth-conf/int
+disparity since both utilize TThreadPoolServer. It was certainly unexpected that the
+addition of wire encryption would yield a faster median run duration. This result
+prompted, as a sanity check, sniffing the net traffic (in a contrived example
+not during a timed run) in both auth-conf and auth-int to ensure that the message
+contents were actually obfuscated in auth-conf (they were) and not obfuscated in
+auth-int (they weren\u2019t).</p>
+
+<h2 id="future-work">Future Work</h2>
+
+<p>Part 2 of this series will consist of the same continuous ingest loads and
+configurations with the addition of a query load on the system.</p>
+
+
+
+<p><strong>View all posts in the <a href="/news">news archive</a></strong></p>
+
+ </div>
+
+
+<footer>
+
+ <p><a href="https://www.apache.org"><img src="/images/feather-small.gif" alt="Apache Software Foundation" id="asf-logo" height="100" /></a></p>
+
+ <p>Copyright � 2011-2017 The Apache Software Foundation. Licensed under the <a href="https://www.apache.org/licenses/LICENSE-2.0">Apache�License,�Version�2.0</a>.</p>
+
+</footer>
+
+
+ </div>
+ </div>
+ </div>
+</body>
+</html>
http://git-wip-us.apache.org/repos/asf/accumulo-website/blob/803c95d0/feed.xml
----------------------------------------------------------------------
diff --git a/feed.xml b/feed.xml
index 6b10896..db56da6 100644
--- a/feed.xml
+++ b/feed.xml
@@ -6,8 +6,8 @@
</description>
<link>https://accumulo.apache.org/</link>
<atom:link href="https://accumulo.apache.org/feed.xml" rel="self" type="application/rss+xml"/>
- <pubDate>Mon, 27 Feb 2017 11:22:59 -0500</pubDate>
- <lastBuildDate>Mon, 27 Feb 2017 11:22:59 -0500</lastBuildDate>
+ <pubDate>Mon, 06 Mar 2017 17:58:36 -0500</pubDate>
+ <lastBuildDate>Mon, 06 Mar 2017 17:58:36 -0500</lastBuildDate>
<generator>Jekyll v3.3.1</generator>
<item>
@@ -182,6 +182,295 @@ HDFS High-Availability instances, forcing NameNode failover.</p>
</item>
<item>
+ <title>Security Performance Implications</title>
+ <description>
+<p>The purpose of this two part series was to measure the performance impact of
+various security configurations on a cluster running Apache Accumulo\u2019s
+continuous ingest suite. The tests were performed using Amazon Web
+Services (AWS), Hortonworks Data Platform 2.4 and Accumulo 1.7. Each of
+the five different security settings in Accumulo 1.7 was tested including
+no security, SSL, and SASL with Kerberos authentication for the three quality
+of protection levels (auth, auth-int, auth-conf). KDC was MIT. HDFS was
+configured to use Kerberos for authentication and had service level
+authorization on. Other than that, no other security settings (HTTPS, RPC
+protection, data transfer encryption, etc) were enabled. Timely was a
+separate, single node HDFS/Zookeeper/Accumulo instance.</p>
+
+<h2 id="intro">Intro</h2>
+
+<p>All runs utilized the continuous ingest suite that ships with Accumulo (a
+standard method to measure performance in Accumulo). It generates random
+graph data and inserts it into Accumulo, creating
+a long linked list of entries. Part 1 was run with just continuous ingest.<br />
+Based on the test results, there was a measurable performance impact as each additional security configuration was put in place.</p>
+
+<h2 id="methodology">Methodology</h2>
+
+<p>We ran 5 tests, one for each security configuration. Each iteration of each test inserted 2 billion entries. Batch writers were configured with 500K max mem
+to artificially inflate the overall write overhead. This was performed on a
+small cluster on AWS.</p>
+
+<p>Each test used one of the following security configurations:</p>
+
+<ul>
+ <li>No security - Default</li>
+ <li>Two way SSL</li>
+ <li>Kerberos/SASL with auth
+ <ul>
+ <li>auth is just Kerberos authentication between client and server. Each end of the RPC definitively knows who the other is.</li>
+ </ul>
+ </li>
+ <li>Kerberos/SASL with auth-int
+ <ul>
+ <li>Builds on auth, also providing message integrity checks of the data going across the wire. You also know that the message you received was not altered.</li>
+ </ul>
+ </li>
+ <li>Kerberos/SASL with auth-conf
+ <ul>
+ <li>Builds on auth-int, also providing confidentiality of the message that was sent to prevent others from reading it (aka wire-encryption).</li>
+ </ul>
+ </li>
+</ul>
+
+<p>For each test, five iterations were run to obtain a min, max, and median
+time elapsed at each security configuration. After each iteration,
+Hadoop, and Zookeeper processes were restarted, Accumulo tables are
+wiped clean and tables are recreated. In addition, pagecache, dentries
+and inodes are dropped by issuing a \u20183\u2019 command on
+/proc/sys/vm/drop_caches to ensure that the OS is not caching things to disk
+that might affect the benchmark. The following sequence was performed
+between iterations:</p>
+
+<ol>
+ <li>Bring down Accumulo</li>
+ <li>Bring down Zookeeper</li>
+ <li>Bring down Hadoop</li>
+ <li>Run sync command</li>
+ <li>Drop OS cache</li>
+ <li>Bring up Hadoop</li>
+ <li>Bring up Zookeeper</li>
+ <li>Bring up Accumulo</li>
+ <li>Drop tables</li>
+ <li>Create tables</li>
+</ol>
+
+<p>For each iteration, the results were stored, fed into <a href="https://nationalsecurityagency.github.io/timely/">Timely</a>, and viewed with Grafana.
+Since the runs were executed sequentially, the start epochs for each run did not align.
+To mitigate, the entries for each run were inserted
+with the same relative epoch for convenient comparison in Grafana.</p>
+
+<p>The table configurations for Accumulo remain the same throughout the
+different iterations and security levels. The Accumulo site
+configurations differ only due to the different settings for the
+security level configurations.</p>
+
+<h2 id="environment">Environment</h2>
+
+<p>In order to perform the testing, a small AWS cluster was setup using 14
+hosts on EC2. Two i2.xlarge instances were used as master nodes and eight
+d2.xlarge instances were used for workers. In addition, two c4.4xlarge
+instances were used for ingesters, one m4.2xlarge instance was used for
+Timely, and one m4.xlarge instance was used for Apache Ambari. A logical
+diagram of the setup is depicted below:</p>
+
+<p><img src="/images/blog/201702_security/figure1.png" alt="" width="400px" /></p>
+
+<p>Figure 1 - Cluster Layout, Roles, and Instance Types on AWS.</p>
+
+<p>The types of nodes and their function are given below:</p>
+
+<table id="instance_types" class="table">
+ <thead>
+ <tr>
+ <th style="text-align: left">Node Type</th>
+ <th style="text-align: left">AWS EC2 Type</th>
+ <th style="text-align: left">EC2 Type Details</th>
+ <th style="text-align: left">Quantity</th>
+ </tr>
+ </thead>
+ <tbody>
+ <tr>
+ <td style="text-align: left">Ingest Nodes</td>
+ <td style="text-align: left">c4.4xlarge</td>
+ <td style="text-align: left">16 core, 30 GB RAM</td>
+ <td style="text-align: left">2</td>
+ </tr>
+ <tr>
+ <td style="text-align: left">Worker Node</td>
+ <td style="text-align: left">d2.xlarge</td>
+ <td style="text-align: left">4 cores, 30.5 GB RAM, 3x2T GB HD</td>
+ <td style="text-align: left">8</td>
+ </tr>
+ <tr>
+ <td style="text-align: left">Master Node</td>
+ <td style="text-align: left">i2.xlarge</td>
+ <td style="text-align: left">4 cores, 30.5 GB RAM, 1x800GB SSD</td>
+ <td style="text-align: left">2</td>
+ </tr>
+ <tr>
+ <td style="text-align: left">Admin Node</td>
+ <td style="text-align: left">m4.xlarge</td>
+ <td style="text-align: left">4 cores, 16 GB RAM</td>
+ <td style="text-align: left">1</td>
+ </tr>
+ <tr>
+ <td style="text-align: left">Timely Node</td>
+ <td style="text-align: left">m4.2xlarge</td>
+ <td style="text-align: left">8 cores, 32 GB RAM</td>
+ <td style="text-align: left">1</td>
+ </tr>
+ </tbody>
+</table>
+
+<p>Table 1 \u2013 AWS Instance Types, Role, Details, and Quantities</p>
+
+<h2 id="results">Results</h2>
+
+<p>The median, max, and min of the milliseconds elapsed
+time of all iterations for each test is displayed below. The percentage change
+columns compare the Median, Max, and Min respectively from the no
+security level to each security configuration (e.g. no security Median
+vs. auth-int Median, no security Max vs. auth-int Max).</p>
+
+<table id="results" class="table">
+ <thead>
+ <tr>
+ <th>Security Level</th>
+ <th style="text-align: right">Median</th>
+ <th style="text-align: right">Standard Deviation</th>
+ <th style="text-align: right">Max</th>
+ <th style="text-align: right">Min</th>
+ <th style="text-align: right">% Change (nosec Median vs. Median)</th>
+ <th style="text-align: right">% Change (nosec Max vs. Max)</th>
+ <th style="text-align: right">% Change (nosec Min vs. Min)</th>
+ <th style="text-align: right">Delta from Previous Level (Median)</th>
+ </tr>
+ </thead>
+ <tbody>
+ <tr>
+ <td>no security</td>
+ <td style="text-align: right">7829394</td>
+ <td style="text-align: right">139340</td>
+ <td style="text-align: right">8143035</td>
+ <td style="text-align: right">7764309</td>
+ <td style="text-align: right">0.00%</td>
+ <td style="text-align: right">0.00%</td>
+ <td style="text-align: right">0.00%</td>
+ <td style="text-align: right">0.00%</td>
+ </tr>
+ <tr>
+ <td>ssl</td>
+ <td style="text-align: right">8292760</td>
+ <td style="text-align: right">87012</td>
+ <td style="text-align: right">8464060</td>
+ <td style="text-align: right">8204955</td>
+ <td style="text-align: right">5.92%</td>
+ <td style="text-align: right">3.94%</td>
+ <td style="text-align: right">5.68%</td>
+ <td style="text-align: right">5.92%</td>
+ </tr>
+ <tr>
+ <td>auth</td>
+ <td style="text-align: right">8859552</td>
+ <td style="text-align: right">134109</td>
+ <td style="text-align: right">9047971</td>
+ <td style="text-align: right">8657618</td>
+ <td style="text-align: right">13.16%</td>
+ <td style="text-align: right">11.11%</td>
+ <td style="text-align: right">11.51%</td>
+ <td style="text-align: right">6.83%</td>
+ </tr>
+ <tr>
+ <td>auth-int</td>
+ <td style="text-align: right">9500737</td>
+ <td style="text-align: right">155968</td>
+ <td style="text-align: right">9753424</td>
+ <td style="text-align: right">9282371</td>
+ <td style="text-align: right">21.34%</td>
+ <td style="text-align: right">19.78%</td>
+ <td style="text-align: right">19.55%</td>
+ <td style="text-align: right">7.24%</td>
+ </tr>
+ <tr>
+ <td>auth-conf</td>
+ <td style="text-align: right">9479635</td>
+ <td style="text-align: right">170823</td>
+ <td style="text-align: right">9776580</td>
+ <td style="text-align: right">9282189</td>
+ <td style="text-align: right">21.08%</td>
+ <td style="text-align: right">20.06%</td>
+ <td style="text-align: right">19.55%</td>
+ <td style="text-align: right">-0.22%</td>
+ </tr>
+ </tbody>
+</table>
+
+<p>Table 2 \u2013 Summarized Time Elapsed for Each Security Level</p>
+
+<h2 id="plots">Plots</h2>
+
+<p>Below are some snapshots of *stats.out elements via Grafana that were inserted
+into Timely with the same relative start time. Each graph represents a field
+in the output generated by <a href="https://github.com/apache/accumulo/blob/1.7/test/src/main/java/org/apache/accumulo/test/continuous/ContinuousStatsCollector.java">ContinuousStatsCollector</a></p>
+
+<h3 id="tablerecshttpsgithubcomapacheaccumuloblob17coresrcmainjavaorgapacheaccumulocoremasterthrifttableinfojaval73"><a href="https://github.com/apache/accumulo/blob/1.7/core/src/main/java/org/apache/accumulo/core/master/thrift/TableInfo.java#L73">TABLE_RECS</a></h3>
+<p>(Number of records in the continuous ingest table. Down sample=1m, aggregate=avg)</p>
+
+<p><a href="/images/blog/201702_security/tableRecs.png"><img src="/images/blog/201702_security/tableRecs.png" alt="" width="800px" /></a></p>
+
+<h3 id="totalingesthttpsgithubcomapacheaccumuloblob17coresrcmainjavaorgapacheaccumulocoremasterthrifttableinfojaval77"><a href="https://github.com/apache/accumulo/blob/1.7/core/src/main/java/org/apache/accumulo/core/master/thrift/TableInfo.java#L77">TOTAL_INGEST</a></h3>
+<p>(Ingest rate for Accumulo instance. Down sample=5m, aggregate=avg)</p>
+
+<p><a href="/images/blog/201702_security/totalIngest.png"><img src="/images/blog/201702_security/totalIngest.png" alt="" width="800px" /></a></p>
+
+<h3 id="avgfilestablethttpsgithubcomapacheaccumuloblob17coresrcmainjavaorgapacheaccumulocoreutilstatjaval63"><a href="https://github.com/apache/accumulo/blob/1.7/core/src/main/java/org/apache/accumulo/core/util/Stat.java#L63">AVG_FILES/TABLET</a></h3>
+<p>(Average number of files per Accumulo tablet. Down sample=1m, aggregate=avg)</p>
+
+<p><a href="/images/blog/201702_security/avgFilesTab.png"><img src="/images/blog/201702_security/avgFilesTab.png" alt="" width="800px" /></a></p>
+
+<h3 id="accumulofileshttpsgithubcomapacheaccumuloblob17testsrcmainjavaorgapacheaccumulotestcontinuouscontinuousstatscollectorjaval127"><a href="https://github.com/apache/accumulo/blob/1.7/test/src/main/java/org/apache/accumulo/test/continuous/ContinuousStatsCollector.java#L127">ACCUMULO_FILES</a></h3>
+<p>(Total number of files for Accumulo. Down sample=1m, aggregate=avg)</p>
+
+<p><a href="/images/blog/201702_security/accumuloFiles.png"><img src="/images/blog/201702_security/accumuloFiles.png" alt="" width="800px" /></a></p>
+
+<p>As can be seen in the plots above, the different security settings have
+relatively consistent, discernable median run characteristics. The big
+dip in each TOTAL_INGEST coincides with a large number of major
+compactions, a rate decrease for TABLE_RECS, and a decrease in
+AVG_FILES/TABLET.</p>
+
+<h2 id="final-thoughts">Final Thoughts</h2>
+
+<p>The biggest performance
+hits to run duration median (compared to default security) were ~21% for
+auth-int and auth-conf. Interesting to note that SSL\u2019s median run duration was
+lower than all SASL configs and that auth-conf\u2019s was lower than auth-int.
+Initial speculation for these oddities revolved around the
+<a href="https://github.com/m1ch1/mapkeeper/wiki/Thrift-Java-Servers-Compared">Thrift server</a>
+implementations, but the Thrift differences will not explain the auth-conf/int
+disparity since both utilize TThreadPoolServer. It was certainly unexpected that the
+addition of wire encryption would yield a faster median run duration. This result
+prompted, as a sanity check, sniffing the net traffic (in a contrived example
+not during a timed run) in both auth-conf and auth-int to ensure that the message
+contents were actually obfuscated in auth-conf (they were) and not obfuscated in
+auth-int (they weren\u2019t).</p>
+
+<h2 id="future-work">Future Work</h2>
+
+<p>Part 2 of this series will consist of the same continuous ingest loads and
+configurations with the addition of a query load on the system.</p>
+
+</description>
+ <pubDate>Thu, 23 Feb 2017 00:00:00 -0500</pubDate>
+ <link>https://accumulo.apache.org/blog/2017/02/23/security-performance-implications.html</link>
+ <guid isPermaLink="true">https://accumulo.apache.org/blog/2017/02/23/security-performance-implications.html</guid>
+
+
+ <category>blog</category>
+
+ </item>
+
+ <item>
<title>Running Accumulo on Fedora 25</title>
<description><p>Apache Accumulo has been available in <a href="https://getfedora.org/">Fedora</a> since F20. Recently, the Fedora
packages have been updated to Accumulo version <code class="highlighter-rouge">1.6.6</code> and have made some
@@ -1725,97 +2014,5 @@ HDFS High-Availability instances, forcing NameNode failover.</p>
</item>
- <item>
- <title>Apache Accumulo 1.6.4</title>
- <description><p>Apache Accumulo 1.6.4 is a maintenance release on the 1.6 version branch.
-This release contains changes from 21 issues, comprised of bug-fixes,
-performance improvements and better test cases. See <a href="https://issues.apache.org/jira/browse/ACCUMULO/fixforversion/12332840">JIRA</a> for a
-complete list.</p>
-
-<p>Below are resources for this release:</p>
-
-<ul>
- <li><a href="/1.6/accumulo_user_manual.html">User Manual</a></li>
- <li><a href="/1.6/apidocs">Javadocs</a></li>
- <li><a href="/1.6/examples">Examples</a></li>
-</ul>
-
-<p>Users of any previous 1.6.x release are strongly encouraged to update as soon as
-possible to benefit from the improvements with very little concern in change
-of underlying functionality. Users of 1.4 or 1.5 that are seeking to upgrade
-to 1.6 should consider 1.6.4 as a starting point.</p>
-
-<h2 id="silent-data-loss-via-bulk-imported-files">Silent data-loss via bulk imported files</h2>
-
-<p>A user recently reported that a simple bulk-import application would occasionally
-lose some records. Through investigation, it was found that when bulk imports into
-a table failed the initial assignment, the logic that automatically retries the
-imports was incorrectly choosing the tablets to import the files into. <a href="https://issues.apache.org/jira/browse/ACCUMULO-3967">ACCUMULO-3967</a>
-contains more information on the cause and identification of the bug. The data-loss
-condition would only affect entire files. If records from a file exist in Accumulo,
-it is still guaranteed that all records within that imported file were successful.</p>
-
-<p>As such, users who have bulk import applications using previous versions of Accumulo
-should verify that all of their data was correctly ingested into Accumulo and
-immediately update to Accumulo 1.6.4.</p>
-
-<h2 id="other-bug-fixes">Other bug fixes</h2>
-
-<ul>
- <li><a href="https://issues.apache.org/jira/browse/ACCUMULO-3979">ACCUMULO-3979</a> Fixed an issue where the BulkImporter failed
-with an error message \u201cQUERY_METADATA already started\u201d.</li>
- <li><a href="https://issues.apache.org/jira/browse/ACCUMULO-3965">ACCUMULO-3965</a> The <code class="highlighter-rouge">listscans</code> shell command did not contain
-the <code class="highlighter-rouge">scanId</code> attribute for currently running scans.</li>
- <li><a href="https://issues.apache.org/jira/browse/ACCUMULO-3946">ACCUMULO-3946</a> Verified that all user-facing operations contained
-appropriate audit messages.</li>
- <li><a href="https://issues.apache.org/jira/browse/ACCUMULO-3977">ACCUMULO-3977</a> Isolated scans with Iterators in use incorrectly
-fail around invocation of <code class="highlighter-rouge">deepCopy</code>.</li>
- <li><a href="https://issues.apache.org/jira/browse/ACCUMULO-3905">ACCUMULO-3905</a> RowDeletingIterator functions incorrectly when
-columns are provided by the client. This restores intended functionality without
-the need for a <a href="https://issues.apache.org/jira/browse/ACCUMULO-1801?focusedCommentId=13970204&amp;page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13970204">workaround</a>.</li>
- <li><a href="https://issues.apache.org/jira/browse/ACCUMULO-3959">ACCUMULO-3959</a> <a href="https://issues.apache.org/jira/browse/ACCUMULO-3934">ACCUMULO-3934</a> Multiple documentation
-improvements to <code class="highlighter-rouge">BatchScanner</code>.</li>
-</ul>
-
-<h2 id="testing">Testing</h2>
-
-<p>Each unit and functional test only runs on a single node, while the RandomWalk
-and Continuous Ingest tests run on any number of nodes. <em>Agitation</em> refers to
-randomly restarting Accumulo processes and Hadoop Datanode processes, and, in
-HDFS High-Availability instances, forcing NameNode failover.</p>
-
-<table id="release_notes_testing" class="table">
- <thead>
- <tr>
- <th>OS</th>
- <th>Hadoop</th>
- <th>Nodes</th>
- <th>ZooKeeper</th>
- <th>HDFS HA</th>
- <th>Tests</th>
- </tr>
- </thead>
- <tbody>
- <tr>
- <td>Amazon Linux 2014.09</td>
- <td>2.6.0</td>
- <td>20</td>
- <td>3.4.5</td>
- <td>No</td>
- <td>ContinuousIngest w/ verification w/ and w/o agitation (37B entries)</td>
- </tr>
- </tbody>
-</table>
-
-</description>
- <pubDate>Sat, 03 Oct 2015 00:00:00 -0400</pubDate>
- <link>https://accumulo.apache.org/release/accumulo-1.6.4/</link>
- <guid isPermaLink="true">https://accumulo.apache.org/release/accumulo-1.6.4/</guid>
-
-
- <category>release</category>
-
- </item>
-
</channel>
</rss>
http://git-wip-us.apache.org/repos/asf/accumulo-website/blob/803c95d0/images/blog/201702_security/accumuloFiles.png
----------------------------------------------------------------------
diff --git a/images/blog/201702_security/accumuloFiles.png b/images/blog/201702_security/accumuloFiles.png
new file mode 100644
index 0000000..4c23277
Binary files /dev/null and b/images/blog/201702_security/accumuloFiles.png differ
http://git-wip-us.apache.org/repos/asf/accumulo-website/blob/803c95d0/images/blog/201702_security/avgFilesTab.png
----------------------------------------------------------------------
diff --git a/images/blog/201702_security/avgFilesTab.png b/images/blog/201702_security/avgFilesTab.png
new file mode 100644
index 0000000..8251bca
Binary files /dev/null and b/images/blog/201702_security/avgFilesTab.png differ
http://git-wip-us.apache.org/repos/asf/accumulo-website/blob/803c95d0/images/blog/201702_security/figure1.png
----------------------------------------------------------------------
diff --git a/images/blog/201702_security/figure1.png b/images/blog/201702_security/figure1.png
new file mode 100644
index 0000000..f28715c
Binary files /dev/null and b/images/blog/201702_security/figure1.png differ
http://git-wip-us.apache.org/repos/asf/accumulo-website/blob/803c95d0/images/blog/201702_security/tableRecs.png
----------------------------------------------------------------------
diff --git a/images/blog/201702_security/tableRecs.png b/images/blog/201702_security/tableRecs.png
new file mode 100644
index 0000000..b7db4ee
Binary files /dev/null and b/images/blog/201702_security/tableRecs.png differ
http://git-wip-us.apache.org/repos/asf/accumulo-website/blob/803c95d0/images/blog/201702_security/tableRecsInMem.png
----------------------------------------------------------------------
diff --git a/images/blog/201702_security/tableRecsInMem.png b/images/blog/201702_security/tableRecsInMem.png
new file mode 100644
index 0000000..b480372
Binary files /dev/null and b/images/blog/201702_security/tableRecsInMem.png differ
http://git-wip-us.apache.org/repos/asf/accumulo-website/blob/803c95d0/images/blog/201702_security/totalIngest.png
----------------------------------------------------------------------
diff --git a/images/blog/201702_security/totalIngest.png b/images/blog/201702_security/totalIngest.png
new file mode 100644
index 0000000..3bc3626
Binary files /dev/null and b/images/blog/201702_security/totalIngest.png differ
http://git-wip-us.apache.org/repos/asf/accumulo-website/blob/803c95d0/index.html
----------------------------------------------------------------------
diff --git a/index.html b/index.html
index cef80f8..0ff029d 100644
--- a/index.html
+++ b/index.html
@@ -164,29 +164,29 @@
<div class="row latest-news-item">
<div class="col-sm-12" style="margin-bottom: 5px">
- <span style="font-size: 12px; margin-right: 5px;">Dec 2016</span>
- <a href="/blog/2016/12/19/running-on-fedora-25.html">Running Accumulo on Fedora 25</a>
+ <span style="font-size: 12px; margin-right: 5px;">Feb 2017</span>
+ <a href="/blog/2017/02/23/security-performance-implications.html">Security Performance Implications</a>
</div>
</div>
<div class="row latest-news-item">
<div class="col-sm-12" style="margin-bottom: 5px">
- <span style="font-size: 12px; margin-right: 5px;">Nov 2016</span>
- <a href="/blog/2016/11/16/simpler-scripts-and-config.html">Simpler scripts and configuration coming in Accumulo 2.0.0</a>
+ <span style="font-size: 12px; margin-right: 5px;">Dec 2016</span>
+ <a href="/blog/2016/12/19/running-on-fedora-25.html">Running Accumulo on Fedora 25</a>
</div>
</div>
<div class="row latest-news-item">
<div class="col-sm-12" style="margin-bottom: 5px">
<span style="font-size: 12px; margin-right: 5px;">Nov 2016</span>
- <a href="/blog/2016/11/02/durability-performance.html">Durability Performance Implications</a>
+ <a href="/blog/2016/11/16/simpler-scripts-and-config.html">Simpler scripts and configuration coming in Accumulo 2.0.0</a>
</div>
</div>
<div class="row latest-news-item">
<div class="col-sm-12" style="margin-bottom: 5px">
- <span style="font-size: 12px; margin-right: 5px;">Sep 2016</span>
- <a href="/release/accumulo-1.6.6/">Apache Accumulo 1.6.6</a>
+ <span style="font-size: 12px; margin-right: 5px;">Nov 2016</span>
+ <a href="/blog/2016/11/02/durability-performance.html">Durability Performance Implications</a>
</div>
</div>
http://git-wip-us.apache.org/repos/asf/accumulo-website/blob/803c95d0/news/index.html
----------------------------------------------------------------------
diff --git a/news/index.html b/news/index.html
index bd07ba2..2ffbe5c 100644
--- a/news/index.html
+++ b/news/index.html
@@ -155,6 +155,13 @@
+ <div class="row" style="margin-top: 15px">
+ <div class="col-md-1">Feb 23</div>
+ <div class="col-md-10"><a href="/blog/2017/02/23/security-performance-implications.html">Security Performance Implications</a></div>
+ </div>
+
+
+
<hr />
<h3>2016</h3>