Gartner defines total cost of ownership (TCO) a comprehensive assessment of information technology (IT) or other costs across enterprise boundaries over time. For IT, TCO includes hardware and software acquisition, management and support, communications, end-user expenses and the opportunity cost of downtime, training and other productivity losses.
wiki描述 Technology deployment can include the following as part of TCO:
Computer hardware and programs
Network hardware and software
Server hardware and software
Workstation hardware and software
Installation and integration of hardware and software
Purchasing research
Warranties and licenses
License tracking - compliance
Migration expenses
Risks: susceptibility to vulnerabilities, availability of upgrades, patches and future licensing policies, etc.
Operation expenses
Infrastructure (floor space)
Electricity (for related equipment, cooling, backup power)
Testing costs
Downtime, outage and failure expenses
Diminished performance (i.e. users having to wait, diminished money-making ability)
Security (including breaches, loss of reputation, recovery and prevention)
Backup and recovery process
Technology training
Audit (internal and external)
Insurance
Information technology personnel
Corporate management time
Long term expenses
Replacement
Future upgrade or scalability expenses
Decommissioning
]]>
<p>先看一段<a href="https://www.gartner.com/it-glossary/total-cost-of-ownership-tco">Gartner</a>对于TCO的定义</p>
<blockquote>
<p>Gartner defines total cost of ownership (TCO) a comprehensive assessment of information technology (IT) or other costs across enterprise boundaries over time. For IT, TCO includes hardware and software acquisition, management and support, communications, end-user expenses and the opportunity cost of downtime, training and other productivity losses.</p>
</blockquote>
迟到的2017总结http://blog.decbug.com/2017_summary/2018-03-30T17:00:00.000Z2018-04-23T01:39:07.600Z从12月底就打算写的，但实在是太忙了，一直拖到现在。。。汗 不知不觉间，就到了18年的3月底，实在是不能再拖下去。顺便不能免俗一下，按照阿里财年做一次总结吧。
When a IPC namespace is destroyed(i.e., When the last process that is a member of the namespace terminates), all IPC objects In the namespace are automatical destroyed.
When a network namespace is freed, its physical network devices are moved back to the initial network namespace. But how about the virtual network device?经过试验，虚拟设备会消失
Pinpoint Agent attaches itself to a host application (such as Tomcat) as a java agent to instrument various classes for tracing. When a class marked for tracing is loaded into the JVM, the Agent injects code around pre-defined methods to collect and send trace data to the Collector. In addition to trace data, the agent also collects various information about the host application such as JVM arguments, loaded libraries, CPU usage, Memory Usage and Garbage Collection.
To wire up the agent, pass $AGENT_PATH/pinpoint-bootstrap-$VERSION.jar to the -javaagent JVM argument when running the application:
-javaagent:$AGENT_PATH/pinpoint-bootstrap-$VERSION.jar Additionally, Pinpoint Agent requires 2 command-line arguments in order to identify itself in the distributed system:
-Dpinpoint.agentId - uniquely identifies the application instance in which the agent is running on
-Dpinpoint.applicationName - groups a number of identical application instances as a single service
if (trace.canSampled()) { SpanEventRecorder recorder = trace.traceBlockBegin();
// RPC call trace have to be recorded with a service code in RPC client code range. recorder.recordServiceType(SamplePluginConstants.MY_RPC_CLIENT_SERVICE_TYPE);
// You have to issue a TraceId the receiver of this request will use. TraceId nextId = trace.getTraceId().getNextTraceId();
// Then record it as next span id. recorder.recordNextSpanId(nextId.getSpanId());
// Finally, pass some tracing data to the server. // How to put them in a message is protocol specific. // This example assumes that the target protocol message can include any metadata (like HTTP headers). request.addMetadata(SamplePluginConstants.META_TRANSACTION_ID, nextId.getTransactionId()); request.addMetadata(SamplePluginConstants.META_SPAN_ID, Long.toString(nextId.getSpanId())); request.addMetadata(SamplePluginConstants.META_PARENT_SPAN_ID, Long.toString(nextId.getParentSpanId())); request.addMetadata(SamplePluginConstants.META_PARENT_APPLICATION_TYPE, Short.toString(traceContext.getServerTypeCode())); request.addMetadata(SamplePluginConstants.META_PARENT_APPLICATION_NAME, traceContext.getApplicationName()); request.addMetadata(SamplePluginConstants.META_FLAGS, Short.toString(nextId.getFlags())); } else { // If sampling this transaction is disabled, pass only that infomation to the server. request.addMetadata(SamplePluginConstants.META_DO_NOT_TRACE, "1"); } }
if (throwable == null) { // RPC client have to record end point (server address) String serverAddress = ((ServerAddressGetter) target)._$PREFIX$_getServerAddress(); int port = ((ServerPortGetter) target)._$PREFIX$_getServerPort(); recorder.recordEndPoint(serverAddress + ":" + port);
TargetClass13_Request request = (TargetClass13_Request) arg0; // Optionally, record the destination id (logical name of server. e.g. DB name) recorder.recordDestinationId(request.getNamespace()); recorder.recordAttribute(SamplePluginConstants.MY_RPC_PROCEDURE_ANNOTATION_KEY, request.getProcedure()); recorder.recordAttribute(SamplePluginConstants.MY_RPC_ARGUMENT_ANNOTATION_KEY, request.getArgument()); recorder.recordAttribute(SamplePluginConstants.MY_RPC_RESULT_ANNOTATION_KEY, result); } else { recorder.recordException(throwable); } } finally { trace.traceBlockEnd(); } }
]]>
<p>近日关于云计算安全的问题，闹的沸沸扬扬。某不才，也算是云计算入门人士，便结合自身经历分析一番。</p>
Run a full OpenStack in a LXD containerhttp://blog.decbug.com/openstack_in_lxc/2017-02-25T16:00:00.000Z2018-04-23T01:39:07.608Z鉴于lxc容器特别像虚拟机的特点，rackspace，都有把openstack部署到lxc容器里的实践，恰好lxc也有类似教程，打算自己跑一遍，看看是怎样一种体验
how to run a full OpenStack, using LXD containers instead of VMs and running all of this inside a LXD container (nesting!).
root@first:~# curl aliyun.com <!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN"> <html> <head><title>301 Moved Permanently</title></head> <bodybgcolor="white"> <h1>301 Moved Permanently</h1> <p>The requested resource has been assigned a new permanent URI.</p> <hr/>Powered by Tengine/Aserver</body> </html>
]]>
<p>鉴于lxc容器特别像虚拟机的特点，rackspace，都有把openstack部署到lxc容器里的实践，恰好lxc也有类似教程，打算自己跑一遍，看看是怎样一种体验</p>
<p>how to run a full OpenStack, using LXD containers instead of VMs and running all of this inside a LXD container (nesting!).</p>
cloud-inithttp://blog.decbug.com/cloud-init/2017-02-23T16:00:00.000Z2018-04-23T01:39:07.600Z刚创出来的虚拟机，要如何设置初始密码？如何添加环境变量？如何进行一系列的预置操作？ 答案就是cloud-init。 当前正在搞openstack+lxc，也需要在创容器的时候注入一些信息，比如初始密码，开启SSH密码登录，允许root等等，就顺手记录cloud-init一下。
writethrough writethrough mode is the default caching mode. With caching set to writethrough mode, the host page cache is enabled, but the disk write cache is disabled for the guest. Consequently, this caching mode ensures data integrity even if the applications and storage stack in the guest do not transfer data to permanent storage properly (either through fsync operations or file system barriers). Because the host page cache is enabled in this mode, the read performance for applications running in the guest is generally better. However, the write performance might be reduced because the disk write cache is disabled. 默认模式，认为host page开启，没有用disk cache，会保证数据可靠，写性能会差，读性能会好？
writeback With caching set to writeback mode, both the host page cache and the disk write cache are enabled for the guest. Because of this, the I/O performance for applications running in the guest is good, but the data is not protected in a power failure. As a result, this caching mode is recommended only for temporary data where potential data loss is not a concern. host page和disk cache都开启，性能最好，但在掉电时会有丢数据的风险
none With caching mode set to none, the host page cache is disabled, but the disk write cache is enabled for the guest. In this mode, the write performance in the guest is optimal because write operations bypass the host page cache and go directly to the disk write cache. If the disk write cache is battery-backed, or if the applications or storage stack in the guest transfer data properly (either through fsync operations or file system barriers), then data integrity can be ensured. However, because the host page cache is disabled, the read performance in the guest would not be as good as in the modes where the host page cache is enabled, such as writethrough mode. 只开disk cache，不开host page，号称写性能最强，读一般
unsafe Caching mode of unsafe ignores cache transfer operations completely. As its name implies, this caching mode should be used only for temporary data where data loss is not a concern. This mode can be useful for speeding up guest installations, but you should switch to another caching mode in production environments. 只写到缓存，不落盘，建议只保存临时数据
]]>
<p>go get, but connect to golang.org timeout!!!</p>
一次容器网络问题debughttp://blog.decbug.com/container_network_debug/2017-01-02T17:00:00.000Z2018-04-23T01:39:07.600Z基于neutron做了一个容器网络方案，需要测试性能，用的iperf，但是在测试过程中，tcp/udp无法抵达跨节点的容器中
]]>
<p>调用链分析有三种模式：</p>
<ol>
<li>白盒，zipkin</li>
<li>灰盒，pinpoint，往JVM里注入</li>
<li>黑盒，优点是难度大，精度一般，优点是不用改代码</li>
</ol>
<p>之前分析的sysdig就算是黑盒,<a href="http://blog.decbug.com/2016/10/12/sysdig/">http://blog.decbug.com/2016/10/12/sysdig/</a>，恰好有需求说想不改代码，由容器云平台提供调用链分析。</p>
<p>参考了几篇论文及业内快讯之后，</p>
<ul>
<li>IBM的<a href="http://www.spark.tc/real-time-application-performance-profiling-using-spark/">Real-time Performance Profiling & Analytics for Microservices using Apache </a></li>
<li>MIT的<a href="https://pdos.csail.mit.edu/~athicha/papers/blackboxes:sosp03.pdf">Performance Debugging for Distributed Systems of Black Boxes</a></li>
</ul>
<p>花了一周时间在k8s上做出来了，当然，由于缺乏算法支持，目前只做到了点对点的调用topo及http,memcached的时延，后面有空再补齐MySQL等等。</p>
