Interesting! Database is also serverless!

1

Cloud database

Cloud Computing Era , I believe many people have used cloud computing products , For example, cloud database .

We buy cloud database , Fixed specifications need to be specified , Such as 2 nucleus 4G , Then pay by specification .

There are two questions , First, the specification is not easy to determine , Because the system load is not sure , Whether to use large size or small size ？

Second, if the specification is determined , The specification is too low during operation / Too high , You need to manually send a request for capacity expansion and reduction , so much trouble .

I often think ： If the database can be loaded , Automatically expand and shrink capacity , That would be nice ！

Recently participated in Tencent technology open day Techoday Online activities , There is a special session about cloud native database TDSQL-C, After listening to it at the beginning, I realized ：TDSQL-C This idea has been realized !

Simply speaking ,TDSQL-C It can help us realize automatic database expansion and contraction , And pay according to the usage , If you put it there without it , No charge at all ！

In fact, this is in Cloud Computing Serverless Thought ： elastic 、 Pay by use , No o & M is required .

Use Serverless Products are like using water and electricity , Turn on the switch to use .

TDSQL-C This database product also has Serverless The ability of , How does it work ？ I went to further study its principle .

2

Realization principle

First ,TDSQL-C Separate computing from storage , Only after separation can flexible resource allocation be made to the computing layer （ Expansion shrinkage capacity ）, This is its implementation Serverless The basis of .

Traditional databases are like this （ Take reading and writing separation of multiple databases as an example ）：

TDSQL-C After separating computation and storage , It goes like this ：

secondly , On the database architecture , Monitor the computing layer ：

The system constantly monitors the computing layer , If there is no request , Just pause it , Realized no charge . Wait until the request comes and restore it , Then, dynamically trigger capacity expansion according to the traffic size / Shrinkage capacity , Programmers don't have to worry about details at all .

In this framework ,TDSQL-C It has three very good features ：

Automatic volume expansion and shrinkage 、 Pay by usage and no charge .

3

Automatic volume expansion and shrinkage

This feature is particularly friendly for companies , Because the flow and load are always variable , For example, the load is stable during the day , Yes CPU And memory usage is relatively stable , In the evening , Need to run some slow queries , Or generate reports , The demand for computational power came up at once .

If we buy large databases , Waste occurs during the day , Buy a small database , Not enough at night , You need to carefully balance cost and performance , It's very annoying .

TDSQL-C It's very thoughtful for us to designate one CPU And memory overview range , For example, the minimum specification 1 nucleus 2G, The largest size 4 nucleus 8G, We don't care about the rest , It will dynamically expand and shrink according to your computing power requirements .

The expansion and contraction speed is very fast , Seconds to complete , We can hardly perceive .

however , It's easy for us to think of a question , The data specification is uncertain , How about the charge ？

4

Charge by usage

The answer, of course, is to charge by usage , The specific method is every 5 Every second CPU And memory consumption , And then according to CCU(TDSQL-C Compute Unit) To charge , Pay the bill every hour .

This CCU The calculation of is also very interesting , Formula is ：

CCU = max(CPU, Memory /2, Minimum size ）

For example, I see , For example, the specification range specified when you purchase is [1 nucleus 2G, 4 nucleus 8G] , The minimum specification in the formula is 1.

Compared with most cloud databases, which charge by volume according to the largest specification in an hour , It can be said that it has achieved a very refined billing method , It really saves money for our users .

There is a more intimate point , The system realizes the resource billing of any unit , The user used 0.4 Then press 0.4 To charge , It's not mandatory “ Less than one core is charged by one core ”.

5

No use, no charge

If our database has no connection for a period of time , Is there any charge at this time ？

For conventional cloud databases , After buying , No use , It's all for a fee .

however TDSQL-C It realizes the separation of calculation and storage , The computing layer can be automatically started and stopped , The starting speed can reach the second level , There are more things that can be done .

If our database instance 10 No connection within minutes , Just pause the instance , Recycle calculation node , At this time, there will be no charge .

When the user request comes , You can quickly wake up the instance , Provide services .

This feature is especially friendly to some low-frequency websites , For example, personal blog , No one visits most of the time , Why pay for cloud database ？ Pause the database instance when no one accesses , Save money without paying . Someone has visited , Start the instance immediately , Pay as much as you want , This is really cool .

There are also some company test environments , At lunch , When I go to bed at night , And weekends are not used , Why pay for it ？ Auto pause , How nice to save some money .

6

summary

TDSQL-C The database is also endowed with Serverless The ability of , It can be like function calculation , Automatic volume expansion and shrinkage , Use more and charge more , Use less and charge less , There is no charge , It has extremely high flexibility and accurate billing ability , It represents the development direction of cloud database .

I strongly suggest that you have a chance to try TDSQL-C, Feel this low cost 、 Very flexible cloud database services .

7

At the end

I participated in this Tencent Techo Day, There are many activities , except “ Use it like tap water ” Cloud native database , The activity also shared Tencent's Lighthouse、Serverless、 Star technology principles such as micro build 、 Applicable scenarios and usage .

From flexible and easy to use TDSQL-C, Out of the box Lighthouse, Low threshold AI Development ...... I can feel the concept of Tencent cloud ： Provide developers with “ Change numerous for brief , Light and easy to use ” Tools and products of , Lower the development threshold , Improve R & D efficiency , Let cloud services benefit more people .

Finally, it is sorted into 《 Tencent cloud lightweight tool guide 》 Courseware materials , Students in need can click “ Read the original ” Or long press the QR code below to download , I think we should know the technical routes of various manufacturers , It's still very helpful for work , After all, there are only a few large development platforms ～