Behind every piece of information you collect, you can't live without TA

Abstract ： Cloud native database GaussDB(for Redis) It not only improves the service efficiency of reading customers , Make personalized recommendation faster and more stable , It also reduces storage and transformation costs , It has laid a cloud foundation for the future development of the enterprise , Help readers achieve higher quality information touch .

With the development of Internet Information Technology

Personalized recommendation has already been integrated into our life

All kinds of information collected in mobile phones

There is... Behind it TA

As a leading content ecosystem service platform in China , Shanghai Yueke Information Technology Co., Ltd （ abbreviation “ Reading guests ”） Drive operations through data analysis , Provide content ecosystem services on a large scale , And provide advertising technology services based on content scenes , Accurately match content and users with technology , Maximize content revenue .

Readers have strong content services and advertising capabilities , More than ten thousand content updates and hundreds of millions of exposures every day . Huge data volume and massive high concurrency , It challenges the database that supports the business application of reading customers .

Storage problems under the surge of data

Database as the cornerstone of carrying massive data , Bear the heavy responsibility of guarding enterprise data assets , It also plays a key role in the digital transformation of enterprises . Data volume surges , What readers use is based on ECS self-built Redis Databases are under great pressure in terms of high concurrency and stability , The cost also rises , The specific pain points are as follows ：

• Performance issues ： In the configuration cache scenario , The reader used Redis Store configuration policy information . There are usually some big key, Big key Open source Redis There are often performance problems in blocking requests , Therefore, the problem of slow query often occurs , There are also a lot of alarms in Yueke's own monitoring group every day .
• Massive data with high concurrent access ： Because the business adopts distributed deployment , Yes Redis There are a lot of concurrent requests , build by oneself sentinel sentry Redis The number of connections is maintained at 3w, Open source Redis Unbearable , This leads to frequent business access timeouts , You even need to restart self built Redis. Again , A lot of alarms are received every day .
• Data storage is expensive ： The amount of data is surging , Bron filters protobuf More and more serialized data , increased TB level . And open source Redis Memory cost pain points 、 Stability pain points begin to appear , Bring certain pressure to business operation .
• Relocation compatibility concerns ： The customer has built two different architectures from the beginning Redis colony , Namely Cluster Clusters and Sentinel colony . Each cluster corresponds to the corresponding client code , And do not support each other . If you choose to go to the cloud , Readers must modify their business code , Then reissue the version 、 go online , The burden of business transformation is large .

Personalized recommendations in the cloud native era

The age of cloud Nativity , Cloud native database based on unified cloud infrastructure , Become the first choice for enterprises to go to the cloud . Yueke follows the development trend of the times , Chose Huawei cloud native database GaussDB(for Redis) As the data base of enterprise digital transformation , It completely replaces the original self built Redis database , Business development has reached a new level .

Performance is remarkable , Content recommendation is faster

Configure the cache business for readers key Performance issues ,GaussDB(for Redis) use Distributed architecture and Multithreading The way of combination , Provides excellent performance , Ensure the continuous and efficient operation of the business . Compared to open source Redis The single thread architecture of , GaussDB(for Redis) The multi-threaded architecture of has more advantages , Even if there is a big key, It will not cause global performance damage . After successful relocation , Readers' own monitoring group responded to the timeout alarm, which was greatly reduced , Configure cache service response in a timely and efficient manner , Content recommendations reach the client faster .

Mass storage , Content recommendation is more stable

GaussDB(for Redis) Provide exclusive connection resources , Customers will build their own sentinels Redis Move to 4 node GaussDB(for Redis) After the instance , Business is really exclusive 4w Connect several resources , And are within the appropriate threshold , Very stable operation , It completely solves the problem of connection number of customer reading business , Under the 100 million level flood peak scenario, you can also calmly face , Content recommendation is more stable .

Blum filter business cost savings 80%

GaussDB(for Redis) Adopt the architecture of separation of storage and calculation , You can buy and calculate independently 、 Storage resources , Avoid open source Redis The frequent waste of computing power costs ; It has strong data compression capability , Especially for the... In the bloom filtering scene protobuf Serializing data works wonders , Realized TB Level data to GB Effective compression of level , The release of the 80% The cost of storage , Completely beyond the customer's imagination , It also paves the way for customers' future business growth .

There is no need to transform the application , One click relocation

GaussDB(for Redis) Provide ”Proxy Universal ” Instance type , Compatible with StandAlone client 、Cluster Client and Sentinel client , There is no need to modify the client business code , And it did “ One architecture is fully compatible ”、“ Business relocation 0 reform ”, Completely dispel the concerns of readers about relocation compatibility . Supported by the R & D team , All dozens of self built units were completed in one week Redis, It realizes efficient and smooth senseless migration .

Cloud native database GaussDB(for Redis) It not only improves the service efficiency of reading customers , Make personalized recommendation faster and more stable , It also reduces storage and transformation costs , It has laid a cloud foundation for the future development of the enterprise , Help readers achieve higher quality information touch .

 

Click to follow , The first time to learn about Huawei's new cloud technology ~