Zheng document keywords based on Ma Haiyuan and Pao:memcached , Redis, distributed cache, console, deserialization, JavaThis document applies to: development and operations staff outline:
- How do I see the serialized data in the cache?
- Bulk Delete to a
- It is necessary to monitor the access of each cache port
- Discache solves these problems.
The distributed cache for e-commerce systems is typically Redis and memcached clusters, with many instances on each node because a business type corresponds to a port and is split very clearly. Since a lot of nodes, a lot of ports, the business is also changing, there are changes at any time, how to manage it? Of course, the most pressing requirement is the following, which will be troublesome if not solved.
0x01. How do I see the serialized data in the cache? Java Engineering typically
serialization of class instance objects stored in cache。 Second, the class objects that exist on different ports are not the same. Finally, there is a version of the class object, and you don't necessarily have the correct online version class file to deserialize. In short, when you want to see what is stored in the key value, it can be very troublesome, so we urgently need a system.
0x02. Bulk Delete to aSometimes on-line cache data want to press a rule to delete a batch, there is no tool to take advantage of the hand, that encounter Yu Ying thing is really rush, especially sometimes on-line preheating data preheating wrong time. If can be in accordance with a wildcard rules batch delete, good Great Yan.
0x03. It is necessary to monitor the access of each cache port.
Slow Log: Redis supports a variety of data structures and also supports fuzzy queries, so it's easy to have slow queries online. Therefore, there is a slow-check monitoring is still very able to find the problem.
Execution of Instructions: Cache instruction processing speed, memory usage, DB Size, and so on, these can be observed in real time is also the best.
0x04.discache solves these problems.2014, in the efforts of Zhang Xin and Pao, Discache released as scheduled, also included in the Idcenter system.
What is Discache? The distributed cache management platform of the litters master station is
node management, data queries, basic information, and slow logsFour parts of the function. The entrance is here: Figure 1 Idcenter main interface Discache The main interface length: Figure 2 discache Node management interface you can search for node name by keyword: Figure 3 Search all distributed cache nodes by order keyword, let's look at slow logs. When the corresponding node is found, click the Yellow action button. Figure 4 A slow log slow log list for a node can also be searched. Why is it so slow in time? These requests are redis pattern-matched instructions, and the response time is indeed much more than the get single key in the case of large data volumes. Take a look at the basic information, which is real-time monitoring: Figure 5 Real-time monitoring information for a node when a node is selected, the page keeps updating the data.
Discache How to solve the most fundamental requirements: deserializing key valuesSelect a cache node, enter a key name, support wildcard query, such as "*index*", "g_*" and so on, if someone has already uploaded the CLASS file, then click on a key can directly display the value content, otherwise you need to upload a CLASS file to deserialize. Figure 6 Key fuzzy search and deserialization as shown, also shows the expiration time of a key. It also supports the Gzip decompression feature, which supports a variety of deserialization methods, as shown in. Figure 7 Serialization drop-down list note that you can click the "Delete all" button to delete the key values you have searched for. Well, we've shown that key-value queries, bulk deletions, and single deletions, slow logs, and real-time monitoring of a distributed cache node. Discache greatly improves the efficiency of cache management in litters projects. Here's how the query works: Convert the cache query to a memacache or Redis set of commands to achieve, Redis has a variety of data structures, then each storage structure of the query instructions are different, interface design only need to enter the specified key value, in the underlying Java In the implementation, various query instructions are attempted until the correct value is queried. -eof-
#研发解决方案 #discache-Distributed cache Query and management system
