Google Cloud challenges AWS with new open-source integrations | Hacker News Reader

Google Cloud challenges AWS with new open-source integrations

techcrunch.com

96 pointspauldix23 comments

Great... how long till google gets bored of this and cuts all theses services leaving their customer's screwed?

After seeing how google handles products and their sun-setting I would never trust my production environment with google.

The only product I ever integrated our production environment with was google maps/places and we got fucked over with a 5x cost increase.

Seriously who would ever trust google over AWS or Azure...

You’re not wrong. I see most cloud consumers at AWS, with anyone else married to Microsoft or who is in business competing against Amazon going with Azure. Which is a shame, because Google Cloud is pretty solid, but the stigma is real.

Is it really a stigma if it's backed by evidence? And repeated often?

https://killedbygoogle.com

I didn't stay the stigma was uncalled for. I personally do not use Google services anymore, nor would I recommend anyone else do so.

Maybe it really helps Amazon that their cloud offering has a separate brand and is pretty much distinct of Amazon. From my experience, few people think of Amazon's other products when evaluating AWS.

Similar with Azure, it's closely linked to Microsoft but still a distinct brand.

Maybe it would've helped Google to name their Cloud product with a name that doesn't contain Google.

Separate like Amazon Web Services and Microsoft Azure you mean?

I find this really strange. This would hardly be a subtle deception. Are people not able to compartmentalize? Commercial is not consumer. GCP is commercial and has SLA‘s and deprecation policies no different from AWS and Azure.

I can't remember when I last heard someone saying "Amazon Web Services". It's commonly called AWS and while most people know it's owned by Amazon, it's a very distinct product.

Google Cloud follows the same naming pattern as other Google products, many of which were discontinued.

These things shouldn't make a difference but I believe they do, and they help to shield AWS from negative press around Amazon.

Amazon and Microsoft don't have the same history of shutdowns on the consumer side though. They both have product churn but Amazon is ecommerce that has always expanded while Microsoft has been providing Windows and Office for decades.

Google makes way more consumer product launches and then suffers from the reputation of shutting these down. So yes, developers should know the difference between commercial and consumer but unfortunately there's too much historical baggage. Also, while the commercial side could be defended before, the Maps API pricing change was just as bad as a shutdown for many customers.

Google Cloud has SLAs with reasonable depreciation schedules. Google's free consumer services, not so much...

This could be a game changer if it is well integrated. No more dealing with forks of open source projects that respect the interface up to a certain point. That way we can a full local stack that is exactly what runs on the cloud.

This looks like a great move - I just wish Citus or another top-tier Postgres option was included, Google's managed Postgres is pretty behind (old version, no PITR, and IIRC limited module and replication support).

We're still managing our Postgres manually ontop of GCE and I'd rather not have to do that, but I also don't want to give up the replication setup we have and the modules we use.

Nice move. GCP is years behind the other clouds in offering common managed services so this seems like a great way to catch up quickly, while allowing the vendors to control their own service.

I'm always surprised at how many of these vendors don't have any cloud offering of their own, or have very limited versions.

Traditionally, AWS has "stolen the thunder" of these vendors, providing a wireline compatible offering (Elastic, Redis, Memcache, Mongo, etc). It's smart of Google to partner with them to help bolster their cloud offering against AWS ("the enemy of my enemy is my friend").

True, although I guess we'll have to wait and see what this means for the operational experience. I cant fault AWS for looking out for its customers first.

I'm very interested to see how this goes. I think Google will be happy they decided to partner with the actual services rather than making their own clones like Amazon is doing. They could stand to gain a lot by offering the "real deal" instead of shitty half finished clones like documentdb.

Google can also use this as PR by propping themselves as the OS friendly cloud who gave these OS companies an olive branch.

If those vendors had their own cloud offering, it would be hosted underneath with GCloud/AWS/Azure in reality. The CAPEX needed to roll your own is way too high when compared to Google/Amazon/Microsoft's budget. Even SAP isn't bothering to enter that battle anymore.

Running on the clouds is what I meant, not their own data centers. We want their managed services next to the rest of our infrastructure and sharing VPCs.

The issue is that they don't have much of an offering at all even though the model is clearly proven. MongoDB Atlas is the only one with comprehensive options.

ElasticCo and Mongo (the companies) already offer managed services that run on top of AWS where you can pay for the full versions. How is this different?

Why would I move to GCP instead of just sticking with AWS and paying for managed services directly from the vendor?

And then having to worry about Google losing interest in a year or two?

Unified billing, support and management was the pitch...if I recall correctly. Seemed pretty compelling to me.

You get unified billing through AWS/Amazon marketplace:

"Elasticsearch Service on Elastic Cloud" Sold By: ElasticSearch Inc

https://aws.amazon.com/marketplace/pp/B01N6YCISK?qid=1554910...

MongoDB Atlas Sold By: MongoDB

https://aws.amazon.com/marketplace/pp/B077D557RX?qid=1554910...

They are both managed and supported by their respective companies and billing is through AWS.

It would be interesting to understand how Google and the open source companies are going to share revenue.

What happens to existing Google-managed open source products? Does partnering with Confluent now render pub/sub obsolete?

I'd say that Pubsub has it's own use cases, as it provides different guarantees than Kafka (most importantly in message ordering that enables different performance characteristics).

Reply on news.ycombinator.com