The Dangers of Cloud-based Integration Solutions

Mark Johansson
Founder and Chief Technology Officer

This is Mark’s featured article in the second Missing Link Technology Partner Newsletter

Since 2001, Missing Link Technology has partnered with organizations like yours to provide best in class enterprise software integration and automation solutions. We enable your software to seamlessly and cost-effectively connect to multiple systems, ensuring that no more sales are lost due to poor or lacking integration. We expect you will shop the market to see what your options are. Here are some things to consider if you are thinking of a cloud-based integration offering for your software:

Vendor Lock-In.  There are few if any open standards for how cloud-based integrations work.  The UI’s are nearly always proprietary, and you are not likely to find a tool to let you export your integration in a way that allows it to be portable to another vendor

Lack of advanced controls and extensibility.  Each cloud integration provider has their own user interface that lacks the advanced controls offered by Visual Studio or similar tools.  If you can make your integration work using a limited tool set, great.  Otherwise, you may be stuck with a vendor that does not have the capabilities that you require.

Security and Privacy.  All traffic flows over public networks.  This introduces additional risks and vulnerabilities.

Availability.  What happens if/when your vendor develops some great new functionality to support one of its customers and then releases that new functionality in an update that affects your integration?  At best you see no impact (and no benefit) from that.  At worst, your integration is shut down. An outage on Amazon Web Services in 2017 publicly traded companies up to $150 million dollars.  If your integration is hosted with that vendor, then your integration is vulnerable.

Paying extra and getting less  You use far more data in a cloud-based integration, and you pay for it.  You must use more data due to the additional security around each transaction  – which is required because all traffic is now going across the open internet. Compare that with an on-premise installation where the data packets are generally smaller, require less security (because they are on a private network) and no one is charging you extra for additional data.  On top of that, the data flows faster because it is all local. Your data is trigger-based (optimized polling), and more streamlined; requires less overall data, less overall transactions; higher throughput, and no additional charges for your throughput fulgurations. On-premise is better in every way.

Duplicated records.  Cloud integration systems are sometimes lacking in tracking mechanisms.  You often have to create your own tracking mechanisms or settle for duplicated transactions that come across every call.  In this case, you must come up with your own mechanisms to ignore the duplicates.

Lack of optimization.  Calls are generally not well optimized.  For instance, rather than having a simple and streamlined call to see if there are any new orders, you often must run the integration to bring over all new orders and just deal with an empty result set if there are none.  This is because the “bring over all new orders” call is optimized for transferring orders, not necessarily for asking if there any new orders.  As a result, this call (which you must poll continually) may be bloated and inefficient for discovering if/when there are new orders to integrate,

“We can’t do that”   Some transactional integration is nearly impossible with a poll-based integration (the kind that is common and, in some cases, your only choice with cloud-based integrations).  For example, if you are monitoring for transactions that impact inventory and three of those transactions that change inventory levels occur within a two-minute timeframe and your poll is set to every 3 minutes, you will only see the last transaction, the first two will be missed.  In this case, the answer you may get is that cloud integration just can’t do that. You might as well just set up a batch-based integration to run every hour or every night and just forget about integrating in real-time or near real-time.

Error remediation.  In cloud systems the workflow and remediation systems for errors are often not as sophisticated as they are for normal-case operations.  Since virtually all cloud integration providers focus their main functionality on normal-case operations, you will settle for errors requiring manual retries or other manual tweaking without the ability to configure the system to make those changes and re-attempts automatically.

There is a better way

With the large number of cloud-based integration solutions, it’s surprising that most offer the same thing: mapping from endpoint to endpoint. Why go for a “me too” cloud-based integration when Missing Link can build you what you really want for less money?

With Missing Link Technology you will have a strategic advantage over your competition without the ongoing data fees and without all the disadvantages of having your integration hosted elsewhere.