5 tips for enterprise data security

Date: Oct 25, 2022

Related Tags：1. What Is the Future Like for CIOs?
2. Alibaba’s CTO On Everything You Wanted To Know About AI Ethics

Abstract: Data security has a decisive impact on the survival and development of enterprises. The leakage and destruction of data assets will lead to irreparable economic losses and lack of core competitiveness of enterprises. Often, the vast majority of small and medium-sized enterprises focus on the rapid development of business and ignore them. Importance of data security.

First, the current data security situation faced by enterprises

Data security is one of the most important issues that enterprise CIOs, CTOs, IT administrators and bosses need to consider when choosing to use any IT product. In the current cloud era, which choice is more secure, public cloud, private cloud or IDC, has always been an enterprise manager. one of the factors to consider.

Regarding this issue, in fact, there is a misunderstanding of many people's cognition, that is, only the hardware is their own, and the data inside is controllable by themselves, so it is safe. But in fact, there is a big difference between the data itself and the real object. The data is composed of binary 0 and 1. Whether the data is safe or not cannot be determined by being around, because the leakage or modification of the data does not require any cost at all. It only needs one time. The network transfer is complete.

Secondly, analyzing the reasons behind a security incident is often related to technology, process and human factors. For example, if the technology is improperly selected, and the data is not physically backed up or backed up in different places, it will often cause irreversible effects; in terms of systems and processes, the authority given to a single person is too high, not to mention deliberate destruction, misoperation is also fatal; human factors Including misoperation, as small as crashing a server, as large as deleting the core database, these are all things that happen frequently.

Therefore, whether the business is deployed in its own IDC or a hosted IDC, as long as it is exposed to the public network, it is also a threat. Whether a device is hosted in an IDC or deployed in a public cloud, as long as it is a server with a public network entry, business security requires a lot of resources and effort to ensure and maintain.

Second, the solution, how to avoid the occurrence of data security incidents

There are many principles of data security assurance, and they all understand the truth, but why do many companies choose to build their own database systems, but still ignore data security? In fact, due to factors such as the company's own technical strength, management level, and investment in IT resources, it is actually difficult to implement the data security strategy mentioned above.

Taking the construction of a distributed storage system as an example, the cost of open source and self-built is not small. Using open source solutions, such as Ceph, GlusterFS, etc., to maintain a cluster of dozens of servers requires at least 1-2 senior storage engineers, and To be able to fully grasp the details of the full-stack technology, few people can be found in China, and the risk of data loss is not a small challenge. Similarly, to be able to maintain a set of database production and backup clusters, the requirements for ordinary DBAs are not low. If you choose a product that is deployed by commercial privatization, the investment of hundreds of thousands of dollars is also unbearable for ordinary enterprises.

1. Database security policy

At present, there are two types of enterprise databases: self-built databases and cloud databases. As an important asset of an enterprise, once the database is lost or damaged, the consequences will be unimaginable. How can we make the data more secure? The suggestion here is to make backups whether you build it yourself or use cloud products.

For users who are already using the self-built database, the emergency plan needs to record the detailed steps for recovery through binlog or other backup files, and conduct regular drills to ensure that such a plan can really work when a problem occurs. In addition, it is necessary to have a good third-party professional data recovery company, so as to avoid the ability to restore data from the disk in the event that the backup file is also deleted. For the scenario where the cloud server builds its own server, it is necessary to combine the periodic cloud disk snapshot function provided by the cloud vendor for data recovery. For cloud database scenarios, you don't need to worry too much about data loss, but you need to be proficient in how to roll back databases on the cloud.

Regardless of cost, cloud databases have more obvious advantages in terms of security and cost-effectiveness than self-built databases. Cloud databases can greatly protect business data while simplifying operation and maintenance operations. Combined with cold backup and binlog, cloud database can provide data rollback capability at any point within 7-732 days. In the case of extreme data damage, you can directly use the cloud database rollback function to restore the data to the time before the damage.

2. Snapshots, snapshots, cloud hosts should take regular snapshots

Snapshot refers to the complete copy or mirror of the data set at a certain point in time (the point in time when the copy starts). When the production system data is lost, it can be completely restored to the snapshot point in time through the snapshot, which is an important data disaster recovery. means.

Snapshots are mainly used for disaster recovery and backup of milestones in the production system. By making a fully available copy of the specified cloud disk, the backup is independent of the cloud disk's life cycle. The snapshot includes the data of the hard disk at the point in time when the copy starts, and does not occupy the user's storage space. For Tencent Cloud, snapshots created by users will be stored in object storage in a redundant manner, thereby further ensuring the reliability of backups. Incremental backup of snapshots means that only changed data is saved, which minimizes the time required to create snapshots and saves storage costs.

3. Do a good job in cloud account permission management

Cloud account management permission management can help customers securely manage access rights to resources under Tencent Cloud accounts. Users create, manage and destroy users (groups) through CAM, and use identity management and policy management to control the permissions of other users to use cloud resources, so that the granularity of resource access permissions under cloud accounts is controllable and reduces the risk of misuse or unnecessary operations. Risk of data corruption and loss.

CAM supports clear permissions, security and controllable solutions through the following functions. For example, sub-accounts can be created in the main account, and the sub-accounts can be assigned the management rights of resources under the main account without sharing the relevant identity credentials of the main account.

In addition, different access rights can be granted to different people for different resources. For example, some sub-accounts can be allowed to have the read permission of a COS bucket, while other sub-accounts or the main account can have the write permission of a COS storage object, etc. The resources, access rights, and users here can be packaged in batches to achieve refined rights management.

4. Implement hierarchical management and encryption of important data

At the data security protection level, the whole life cycle protection strategy is shifted from network-centric to data-centric. That is to implement data classification and classification, sort out the data life cycle status, and plan corresponding data protection strategies such as data encryption, desensitization, and auditing according to different data sensitivity levels and data usage status, to ensure that data security is controllable throughout the process. For the core important data that affects business operations, cryptographic technology should be applied in the process of data generation, flow, storage, use and destruction, and resource-level fine-grained identity authentication and access control should be implemented to prevent external internal Business data security risks caused by unauthorized personnel access.

5. Establish data security protection for the entire life cycle

The data life cycle covers multiple stages of data creation, storage, use, sharing, archiving, and destruction. It faces threats from external attacks, internal leaks, and big data sharing. The means of protection against different threats vary widely.

For external attacks, identity authentication, database auditing, and encryption gateways are used to protect core data from external attacks;
For internal data leakage, adopt 4A and DLP and other security capabilities to comprehensively protect the data leakage risk of enterprise operation and maintenance, office, data analysis and other scenarios;
For data leakage in big data sharing, build security capabilities such as desensitization, watermarking, encryption, auditing and authority control.
Therefore, enterprises need to sort out the risk points as a whole, and then carry out overall planning and joint defense. And build different solutions for different scenarios such as external, internal, and big data.

As a unified portal for data operation and maintenance on the cloud, the bastion host has the capabilities of account authority management, password management, and command management. It can eliminate the risk of malicious command execution such as most unauthorized operations and database deletion for enterprises. Due to the centralized management mode, the operation and maintenance personnel must pass the unified authentication of the bastion machine before they can operate the server and database. Therefore, as long as the security policy is set on the bastion machine, it can be easily blocked and the risk of data loss can be greatly reduced.

3. What are the public cloud data security measures

Under the trend of comprehensive cloud adoption, cloud computing manufacturers have accumulated rich experience in data security protection through years of practice, and are exporting them to the public cloud through productization, so that enterprise users can use them out of the box.

1. Cloud hard disk CBS, providing real-time snapshots and second-level recovery capabilities

Tencent Cloud Drive adopts a distributed block storage architecture. Each data block has three copies in the availability zone, which can avoid data damage caused by physical disk and downtime failures. In addition, through the snapshot technology of cloud hard disk, data can be restored to the state within an hour in "second level".

2. Cloud object storage COS, version rollback, remote disaster recovery

Object storage COS can enable the version control function to realize the version control of object storage. After the version control configuration is enabled, the delete operation is equivalent to adding a delete mark; the data of any version in the past can be accessed by specifying the version number, and the data recovery can be realized. Rolling operation to solve the risk of accidental deletion and overwriting of data.

In addition, the object storage also provides the function of cross-region replication, helping users to replicate all incremental files to data centers in other cities through dedicated lines to achieve remote disaster recovery. When the data in the primary bucket is deleted, the data can be restored from the backup bucket by batch copying.

3. ApsaraDB for MySQL provides a flexible and reliable disaster recovery solution for structured data

ApsaraDB for MySQL also implements a wealth of data backup and recovery functions while implementing a highly available service architecture, ensuring that data can be rolled back to any timeline. At the same time, for all high-availability instances, data backup within 7 days and the instance recycle bin retention policy are enabled by default to ensure that data recovery can be guaranteed due to human error. In addition, by configuring a remote disaster recovery instance for the instance, ApsaraDB for MySQL performs data replication in real time, which can easily implement data remote disaster recovery and avoid the risk of data loss caused by regional failures.

4. Tencent cloud data product series, low threshold for security monitoring and auditing

Tencent Cloud data security product series can realize comprehensive monitoring, alarming, post-event auditing and other functions of security events. Tencent Cloud Fortress combines artificial intelligence technology to provide operation and maintenance personnel operation audits for enterprises, alert abnormal behaviors, and prevent internal data leakage.

5. CAM cloud authority management, establish a reasonable authority control system for assets on the cloud

For cloud-based enterprises, account security and reasonable resource authorization are the first door locks to build a three-dimensional protection system. The authorization of resource management on the cloud should avoid the following risks:

Use Tencent Cloud primary account for daily operations
Created a sub-account for an employee, but the authorization is too large
No access condition control for high-privileged sub-account users and high-risk operations
User permissions and login information are not regularly audited
Management systems and processes that lack authority
Here is an example of COS data access control: granting sub-accounts to different teams in the company, and controlling the resource operation scope of different accounts at different levels through access methods and account permission isolation. Permissions for high-risk operations (such as deleting data) can be separated, and only console operations are allowed. At the same time, MFA verification is used for secondary authentication.

Conclusion

By sorting out the data security incidents that have emerged in recent years, but also information sales by internal staff, deletion of databases by ex-employees, misoperation by developers and testers, etc., behind data security incidents caused by various reasons It reflects that it is difficult to achieve the real security protection effect only by relying on a single point of protection, and building a security protection based on the whole life cycle has become an inevitable choice. It is worth mentioning that under the trend of enterprises adopting the cloud, most discussions on data security should start from the cloud environment, and cloud-native data protection technologies and strategies will also become the main means at present and in the future.