3 ms·
You can change the work factor to scale the amount of time it takes to do the processing. Beyond that, there's a cost to security. If you actually need securi
by dpark 15y ago
You can change the work factor to scale the amount of time it takes to do the processing. Beyond that, there's a cost to security. If you actually need security, you may need to pay for it. How much is it going to cost you to buy an extra server vs the loss of business due to a serious security breach?
- falcolas 15y agoIn the court of public image, simply loosing the data in the first place will cost you business, now matter how secure that data may be. The trust to secure your data will be lost, regardless of whether you used bcrypt or sha or plaintext. If you consider the loss of business to be pretty much constant in the event of a loss of control of the customer's data, what is the value of extra server(s) (plus the cost of maintenance) for one of the least valuable pieces of information a customer can give you? As an example, look at Instapaper. How many people are more concerned that the government has access to their sha1 encrypted passwords as opposed being concerned that the government has free access to their reading history?
- dpark 15y agoFor sure there's going to be loss of confidence either way. You're probably going to have a bigger loss if the security experts are pointing out that you couldn't even get password storage right, though. Obviously they It's up to you whether you care enough about your customers to try to get security right. You gave a reasonable argument for why using bcrypt instead of md5 makes little financial sense. However, that same argument applies to plaintext vs hashed. Or hashed vs hashed with salt. In the end, you need to decide how much you care about security, and also weigh how much security you are morally obligated to provide.
- falcolas 15y agoPlaintest vs. hash, or salted hash vs. salted hash do not incur the same performance penalty. And while I may feel morally obligated to provide great security, the people I report to would not be as impressed with that in a cost-benefit analysis. tl;dr It's a great idea, but a tough sell, particularly at scale.
- dpark 15y agoMy hashed/unhashed comparison was not a performance statement, but a trust statement. If a breach will lose the same amount of confidence regardless of password storage, why bother with even hashing? I can't imagine how "good security - scalable enough for Twitter" can be a hard sell. Your customers certainly won't by sympathetic if they learn that you willfully chose weaker security to save a buck.