Big data means exposing more data to hackers?

This seems to have oversimplified the problem of large data. To some extent, managing large data is like taking a child, with two children doing more than twice times the amount of a child, more like an exponential relationship. As large data grows, the potential management problems will grow exponentially.

For large data, you have to understand that it not only means more data, it also means more complex data, more sensitive data, and it also means it may expose more data to attackers who successfully infiltrate the network. If a potential attacker knows you have a large quantity of high-quality data, this may increase your attack surface because you are considered an attractive target. This should be considered when you are conducting an enterprise risk analysis.

But when we talk about big data, we don't just talk about volume or quantity. Most people will talk about velocity, the speed at which data enters the enterprise environment. Personally, I think the concept of produced (kind) is more interesting. What is this data? Managing large amounts of PHI data is much simpler than managing mixed data, including PHI data, PCI data, medical data, and demographic data. The similarity of data has great influence on data management. How similar is the data? The more diverse the information, the more complex the infrastructure used to support this information.

It deployments that support large data are more relevant than data-supporting operations, rather than the security of a database. Scalable infrastructure, parallel processing, data replication, and large amounts of memory processing are just some of the discussions about large data operations. But what does big data security mean for US security personnel?

1. Understanding the Data

You have a lot of data, but do you know the data? Do you have PCI data, PHI data, or private enterprise data/customer data? Before you worry about anything else, you need to know what data is included in the big data. Knowing big data can help you better manage the data and allow you to discover abnormal data. Cleaning up irrelevant or erroneous data is not a job to take lightly.

2. Infrastructure issues

Availability is definitely a security issue, so make sure you have the proper size of the infrastructure. Is your network fast enough to support data throughput requirements? Do you have enough CPU capacity to support the movement and management of data between applications, databases, and storage devices? Do you have enough disk space to store the data? Do you have a strong enough hard drive management program? These are standard it issues to consider in managing large amounts of data. In many ways, the more data you have, the harder it is to protect, and your solution needs to be scaled up as your data grows and your data needs. Can you encrypt petabytes or EB-level data in real time to ensure that the data meets business requirements (including meeting time requirements)?

3. Understanding Timing constraints

Timing is a very important factor. Does your data have a lifecycle? For example, in terms of timing, clinical medical information is clearly more important than typical manufacturing metrics data. More bluntly, some data will lose some value if not managed and analyzed in time. Do you think the system will be of any value if the Phalanx missile defense system takes five minutes to assess the threat and response? The answer is clearly negative. This certainly drives it capacity and throughput requirements, and sometimes timing doesn't matter, but in many cases the old data may be irrelevant and the life cycle of the data is more important than we think.

4. Understanding Content Data

This is a direct extension of the above three questions to help us understand large data. When data has specific content, we can manage it as information, not bytes and bits. Is it PHI data, PCI data, or personal information? We can dig deeper into this data, manage it through content, and not just treat it as "data." Processing this data in an intelligent way can also allow us to work in a similar way with data with similar content, and we can build relationships between those data.

While large data may be just "data", we really don't want them to be "data", we want them to be "information" (that is, data with content). Large data as a source of data analysis is more valuable than mere "data", this is why data content and relevance are so important that we can make data more "intelligent" through data content and relevance, rather than talking about "medical data", but about unique patient identification, allergies, current prescriptions, and more. You are not just talking about "manufacturing data", but about specific inventory items, suppliers, http://www.aliyun.com/zixun/aggregation/31945.html "> Commodity prices, sales prices, buyers, etc." You are not just talking about security event data, but are talking about IDs and internal system reports for attacks against the Modor system (which is a Windows Server 2008 R2 SP1 running Oracle 11g Enterprise that contains all clinical patient information).

From a certain point of view, all this is only exacerbating the problem. We need to deal with a large number of potentially valuable, dynamic, and complex data, and then analyze the data for content. The analysis itself, and the processes used to create these analyses, are also valuable. After all, big data is really just a bunch of data if we can't get the relevant information from the Big data analysis. This also gives us an understanding of the need to protect these analytical modeling and results, as well as access to them.

(Responsible editor: Lu Guang)