What do you think use cases are for hybrid deployment model for API management?

prabhat
Participant V

In the webinar that I did few weeks back (See this 2-minute segment at https://youtu.be/cSeSMQ5sMJc?t=683), I talked about few use cases where Microgateway fits the bill for hybrid cloud API management. I am sure there are more use cases and would love to know your opinion.

Tagging few folks that I know have opinion. 🙂 @Kristopher Kleva @epackwood @TestAppUser @Bryant Olson @mikael.mellgren @Karan Vij

2 14 3,328
14 REPLIES 14

Not applicable

Hi @prabhat - the two uses-case we've seen related to a hybrid solution are related to legal (jurisdictional) and geographic boundaries. Several of our global products (which leverage AWS for North America) must be run in regional data centers (i.e. on prem) to satisfy legal requirements around personal data and commerce, and/or latency issues associated with AWS in certain geographies.

The hybrid solution allows us to manage these products virtually as a single identity despite being deployed in the cloud and on-prem simultaneously.

cc'ing @Birute Awasthi

Thanks Allen.

Not applicable

@prabhat some use cases for hybrid that I can think of off the top of my head are:

Cost management

Disaster recovery

Spare capacity

Market segregation

Testing of infrastructure/scaling/etc.

QOS guarantees for select customers/partners

Traffic shaping, ideally dynamic

Behavioural testing

Increased flexibility on capacity and cost

Workflow/process integration (i.e. easier to fit your own box into existing/new policies and procedures than a cloud hosted one - while this may seem like a pure private cloud point there are additional advantages of the cloud ^^^ that they may want to utilise)

Political objections to a pure cloud play from within the company

I could go on...

Thanks Paul

Not applicable

Hello,

I have an extra scenario.

Companies are moving, more and more, their applications to cloud DDC. Which leave us, with this heterogeneous environment, with private prem DDC application and private cloud DDC application. This transition setup that probably will be "forever", how can we forecast future proof architecture?

Have any of your customers, started to ask how can we have this 2xHybrid setup, meaning; e.g. an api.COMPANYBRAND.com for cloud DDC, and api.internal.COMPANYBRAND.com using different gateways. i.e. for could DDC using apigee edge gateway, with horizontal/vertical scale out of the box and for internal DDC we use microgateway in level 1/DMZ, both use apigee edge for analytcs, SDK etc...

Any recommendations for this kind of scenario and what companies are adopting?

Best regards,

João Paulo

The scenario you explained is the most common use case for Hybrid Cloud API management using Microgateway. There already are couple of customers which use case is what you describe.

Prabhat meaning? On your youtube video, you talk about hybrid but on the "Apigee context" i.e. using prem microgateway and cloud edge. To address the use case you describe on having more prem APIs. What I have described is to use, the same Apigee instance, 2 gateways and 2 domains for a hybrid company aprouch. One gateway is on the cloud Apigee edge and the other the microgateway on prem.

Maybe I got it wrong from the video 🙂 but it looks like is a recommendations of prem Hybrid apigee edge and not both.

//J

I have reread your comments multiple times and I still don't see where we differ. 😉 I am probably making an assumption that you are not making.

You have two different domains one pointing to gateway on the Apigee cloud and other , pointing to on-prem using Microgateway but both gateways are using same Apigee instance i.e Edge for common API management stuff such as analytics, quota and key management.

If you think we are going in circle then we can have a hangout call.

Now we are in the same page 😉 What created my doubts was that in the video or in your previous comments you did not mentioned this scenario of 2 domains using same edge instance from API mgt.

That lead me to an mutual exclusion hybrid architecture recomendation, i.e. with either, Prem+edge API mgt cloud or full cloud Gateway+API mgt. 🙂

But now, I understand that there wasn't mutual exclusion in that sense. and that is also recommended architecture for a company like mine. One thing that made me have concerns was to setup 2 domains for apis one internal and one external and when make the difference of using one or another. even if we have the same API Mgt edge in the cloud which is the most important is still a hybrid hybrid logical setup.

We can have a call later when I am setting this at my corporate

Now i am in Thailand enjoying holidays 😉 Thank you for the help.

Not applicable

Prabhat meaning? On your youtube video, you talk about hybrid but on the "Apigee context" i.e. using prem microgateway and cloud edge. To address the use case you describe on having more prem APIs. What I have described is to use, the same Apigee instance, 2 gateways and 2 domains for a hybrid company aprouch. One gateway is on the cloud Apigee edge and the other the microgateway on prem.

Maybe I got it wrong from the video 🙂 but it looks like is a recommendations of prem Hybrid apigee edge and not both.

//J

Not applicable

Especially in a hybrid environment, it would be very interesting to have Cloud Based Routing, but have the actual Message be processed behind the firewall, so to speak.

Another way to put this, is the idea that your MPs are selectable in the bundle... In this way, the control can be cloud based, monitored and centralized, but the data never has to leave the corporate veil...

Would we not achieve this by conditional routing functionality on Edge where if we detect that request is something to be handled by Microgateway, MP on cloud use Microgateway as target which in turn proxies the real target API?

But in this case, if the call come from your IntrAnet, to consume the API in your IntrAnet you will add the extra hop, i.e. increase latency.even if afterwards the conditional router points to your on-prem load balancer->microgatway-target.

This is the use case we are currently facing. We have a single domain api.<companyname>.com but have the proxied services both onPrem and AWS based.

Is there any best practice for routing the traffic to minimise network hops/latency when accessing APIs from both external internet traffic and internal application traffic?