We’ve been working with pre-2.0 Tyk in dev/staging environments and are considering using it in production.
I have some resistance to recommending Tyk 2.0. It suddenly became rather expensive given our need to have it running in multiple nodes over both production and lower environments.
Given the licensing costs, Tyk Cloud now seems more appealing. I’m not sure it meets our requirements, but if it does, it may be considerably more cost effective.
A few questions:
Does Tyk Cloud with hybrid deployment allow for multiple gateway nodes per account/subscription?
How would you advise configuring multiple gateways (with, for example, mocked responses for only certain environments) per dashboard instance? Is there a difference here between Tyk Cloud and Tyk Professional?
I appreciate your concern, to answer your questions:
Yes, with a cloud plan you can run as many hybrid nodes as you like and manage them from your dashboard account.
You would make use of API Sharding (tagging), Hybrid Agents can make use of our zoning mechanism to operate in zoned (tagged) groups, and mange them all from one dashboard account. So you could have:
Zone 1 (EMEA):
Instance a, b, c (tag: dev env)
instance d, e, f (tag: staging env)
instance h, i, j (tag: prod env)
Redis 1
Zone 2 (US East):
instance a, b, c (tag: staging env)
instance d, e, f (tag: dev env)
instance h, i, j (tag: prod env)
Redis 2
Zone 3 (Australia):
instance a, b, c (tag: staging env)
instance d, e, f (tag: dev env)
instance h, i, j (tag: prod env)
Redis 3
Each of these zones can have their own Redis DB running, they do not need to share one.
Now you can create API Definitions that are all mocks and tag them with “dev”, and as those mocks start being filled in with real endpoints and are ready for QA, you edit their tags and change them to “staging”, as soon as you do this, all Zoned instances that are tagged with “staging” will load up your API Definition - seamlessly transitioning from one environment to another.
If you combine this with service discovery the API definitions could be auto-configuring and DC aware once they load up in their respective tag groups.
In your case you might not need the zones. But groups/zones are really useful for isolated VPCs.
In the above case you have one dashboard, and then multiple Tyk Hybrid node clusters that load up your API definitions as you lifecycle them up and out.
There are some limitations in Tyk Cloud such as certain features not being available (uptime tests and detailed log recording), however otherwise it is the same stack as Tyk Pro, just in a hosted SaaS environment.