distributed

We are faced with the following problem that involves keeping the following in sync: 1 Centralised Server (IIS / MSSQL 2005) Many Desktop WPF clients distributed by ClickOnce Many Mobile clients - (Windows CE) alt text http://img502.imageshack.us/img502/8246/deployment.png With these thorny constraints: all sync relationships are bi-directional the desktop & mobile nodes require offline mode the mobile nodes cannot sync with the central server but are to sync with the desktop nodes over USB. The desktop nodes act as a server for the mobile clients, not the central server. Updates from the

I've read a lot on evenly distributing elements using css but every solution i've found requires you to know and define the width and/or the number of elements being distributed. I have a container that is of a fixed width - within here could be any number of li elements of an automatic width which need to be distributed evenly across the parent element from left to right. Here is my markup - except there could be any number of tabs with any length text in them. <ul class="tabs"> <li class="tab active" id="tab-1"><span class="tab-title">Tab 1</span></li> <li class="tab " id="tab-2"><span class

原文同步至 http://waylau.com/cap-theorem/ 本文介绍了介绍了分布式系统著名的 CAP 理论。什么是 CAP 理论？为什么说 CAP 只能三选二？了解 CAP 对于系统架构又有什么指导意义？本文将一一作答。 什么是 CAP 理论 在计算机科学理论，CAP 定理（也称为 Brewer 定理），是由计算机科学家 Eric Brewer 提出的，即在分布式计算机系统不可能同时提供以下全部三个保证： 一致性（Consistency）：所有节点同一时间看到是相同的数据； 可用性（Availability）：不管是否成功，确保每一个请求都能接收到响应； 分区容错性（Partition tolerance）：系统任意分区后，在网络故障时，仍能操作 为什么说 CAP 只能三选二 下面分别举例说明了为什么说 CAP 只能三选二。 上面的图显示了在一个网络中，N1 和 N2 两个节点。他们都共享数据块 V，其中有一个值 V0 。运行在 N1 的 A 程序可以认为是安全的、无 bug、可预测的和可靠的。运行在 N2 是 B 程序。这个例子中，A 将写入 V 新​值，而 B 从 V 读取值 系统预期执行下面的操作 首先写一个 V 的新​值 V1 然后消息（M）从 N1 更新 V 的拷贝到 N2 现在，从 B 读取将返回 V1 如果网络是分区的，当 N1 到 N2

The best way to do it would be to get the representation of the function (if it can be recovered somehow). Binary serialization is preferred for efficiency reasons. I think there is a way to do it in Clean, because it would be impossible to implement iTask, which relies on that tasks (and so functions) can be saved and continued when the server is running again. This must be important for distributed haskell computations. I'm not looking for parsing haskell code at runtime as described here: Serialization of functions in Haskell . I also need to serialize not just deserialize. Unfortunately,

What are some of the key differences between these two technologies? Does one have obvious advantages over the other? RV is like a radio broadcaster and EMS is like a telephone. If you want to send a message to everyone in town (e.g. the weather forecast for today) then a radio is good because one message goes to everyone simultaneously. Telephone is bad because it takes a long time to call everyone and you pay 20c a call. If you want to tell someone your credit card number you would use the telephone because you can be sure you are talking to the right person. Radio is bad because then

I've read Distributed Tensorflow Doc , and it mentions that in asynchronous training, each replica of the graph has an independent training loop that executes without coordination. From what I understand, if we use parameter-server with data parallelism architecture, it means each worker computes gradients and updates its own weights without caring about other workers updates for distributed training Neural Network. As all weights are shared on parameter server (ps), I think ps still has to coordinate (or aggregate) weight updates from all workers in some way. I wonder how does the aggregation

We are in need of a distributed architecture messaging system/service bus with the capability of Publish/Subscribe. Does anyone have any reccomendations for a framework that we can use for .net applications? Dale Ragan NServiceBus is growing in popularity. It is open source as well. Here is a Hanselminutes episode with Scott Hanselman talking with Udi Dahan about NServiceBus to help grok it. You should definitely evaluate using it. UPDATE: There's also a DNR TV episode which shows what it's like to build an NServiceBus solution from scratch here: http://www.dnrtv.com/default.aspx?showNum=199