When Twitter Falls Down, Do You Stay Up? What Developers Can Learn from Twitter’s Latest Outage
Not even the social media mogul Twitter is immune to outages.
Beginning just after 2am EST on Tuesday, a wide-spread outage gave the world a lengthy opportunity to ask:
What would the world be like without Twitter?
For one, we’d probably all have a lot more time on our hands — but what about for organizations that rely on Twitter’s API for their own website or application?
Here are a few examples of the disruptions that have already taken place:
- Web sites and apps without a backup login process rendered inaccessible
- Poorly designed mobile apps that integrate with Twitter become unresponsive
- Delay in processing integrations to social data, causing a backlog of issues once online again
- Product signups and content traffic driven by social media campaigns dip
- Users can’t get their morning news and can’t message each other
In today’s world, we’re used to things going down, but when it’s a platform as popular as Twitter, the impact can be severe.
Even if you don’t use Twitter yourself, much of the businesses you interact with do — from your local coffee shop to the metropolitan digital advertising agency across the street.
There is also a proliferation of local and state law enforcement, emergency services, and public systems that use Twitter for rapid communication of important information. Police and fire officials (@LOCAL_718) often use social media to alert people, which in turn helps to spread the word during times of emergency.
Is digital interdependence a problem?
Yes and no. Depending on others, such as an API provider like Twitter is critical to shifting focus back onto the software or services you primarily provide. Identity management is a risky business to begin with, so relying on a trusted 3rd party is a given for most businesses.
The trick, however, is to never take anything for granted.
If you integrate with another system, make sure you monitor that system for availability and uptime. You may find that they are in violation of their SLA, and that it’s either time for you to switch to another API provider and/or demand a refund.
Likewise, testing your integrations before expecting real users to use them is crucial to the speed of response when (not if) an issue occurs in your production systems. Testing builds familiarity with how your production services work, not just when they behave as a ‘happy path’, but especially when they don’t behave.
What will your login process look like the next time Twitter goes down for more than a few minutes? How will your revenue stream or release process be affected if this happens to GitHub?
How can I safeguard against being too dependent?
The answer is simple: don’t put all your eggs in one basket.
This recent outage underlines the importance of having more than one authentication method in place before rolling out a product, service, or integration that relies on 3rd party login.
For login processes, make sure that you give users more than one social authentication option. It may take a small amount of additional work upfront, but it will be worth it in a crisis. You may even want to provide your own application-level login mechanism, but just make sure you’re following best practices over storage of credentials and security protocols.
How does a social media outage affect my business?
As a final thought, I challenge you to ask yourself and your team:
“Do we even know how a Twitter outage affects our business?”
Much like Disaster Recovery (DR) planning, simply asking the questions around how your revenue might be affected by a 3rd party system (i.e. Office365, Salesforce, etc.) will help to clear off the dust in your organization and make people aware of dependency risks.
It’s better to know the risk involved that to be blindsided unexpectedly.
How have you been affected in the past by 3rd party API outages or performance problems?
Looking for more advice on API testing? Download our free guide, The Definitive Guide to Data-Driven API Testing for SOAP & REST APIs.