Hacker News Reader

Comment by CyanLite2 | Hacker News Reader

Postgres doesn't come close to the enterprise-level features AND support as that of Oracle or SQL Server.

I’m curious, what features are you missing? Support is always going to be a bit different with open source products, but I’m sure there’s companies that’ll give you a support contract worth more than the equivalent from MS or Oracle.

Out of the box clustering with automatic failover and recovery that "just works". It can be a sticking point when selling product which only uses Postgres to some enterprises. They want to start a few instances, put a load balancer in front of at least a couple of them, and then call it a day.

I think the Citus folks would like a word with you. If you are willing fork over what Oracle asks for, you can probably strike a deal with those folks.

Why would Citus folks like a word with someone wanting something that "just works" "out of the box"? It is reasonable to want/expect this feature to be part of the core DB distribution and handled, documented, maintained the same as other core DB features.

There are commercial solutions that provide that, EnterpriseDB will happily sell you licenses for their solution if you don't want to roll your own.

Distributed transactions, raw filesystems, nice graphical admin tooling with reporting, stored procedure development and debugging capabilities (pgAdmin is nice but that is about it).

A few key missing features:

- You can't write off or expense three martini lunches and golf excursions with a PostgreSQL sales rep.

- When you deal with unavailability, you can't point to the PostgreSQL support SLA's four hour response and tell your boss "not our problem".

- It's harder to justify larger budgets for your IT fiefdom and thus your sense of self-importance when you don't pay for unnecessary license fees.

So I know your comment is a bit tongue in cheek, but there are real concerns about the last two points that people may not realize. At many large orginizations the outside vendor may be the _only_ real technical capabilities that team has so there is no solution other then "outsourcing" to the vendor's support team.

Budgets have similar incentives that make teams want to keep a large budget around even if they could get the job done with less resources. That slush money comes in handy to have the vendor do all sorts of work that isn't really about getting their system running ( i.e. doing all the integration with other parts of your stack )

I think lots of open source products or smaller teams think winning those "Enterprise" deals is about having the better product and checking some boxes on a requirements sheet. When you get into enterprise its all about the relationships

> You can't write off or expense three martini lunches and golf excursions with a PostgreSQL sales rep.

That's one of the Oracle features supported by EnterpriseDB’s proprietary downstream version.

> When you deal with unavailability, you can't point to the PostgreSQL support SLA's four hour response and tell your boss "not our problem".

Also, this, but it is a 24 hour resolution window for Severity 1 issues.

> It's harder to justify larger budgets for your IT fiefdom and thus your sense of self-importance when you don't pay for unnecessary license fees.

EnterpriseDB addresses this, but certainly doesn't usually match Oracle in the budget impact department.

I know your post is sarcastic, but some of these (eg. 2) are real concerns for large enterprises. Enterprises want not just a product but a solution and that includes support and all that comes with it.

You can pay for Postgres support as well.

Sometimes it’s clustering, upgrading servers, replication layouts that work and don’t have random issues with WAL, encryption at rest without requiring the os to do it (has this been fixed?).

Maybe having clustered indexes, good partitioning, etc.

Still, I’m hard pressed to ever recommend anything other than PostgreSQL or MySQL.

What features of Oracle or SQL Server would you say it is lacking?

Regarding support, there's plenty of fantastic companies providing support and consulting services for PostgreSQL and having used plenty of commercial databases I know of know no other database that has as vibrant a community providing amazing free support as well.

SQL Server has a column-oriented index type. Oracle has a thing that lets you query based on what the data was at a particular (recent) timestamp. They both have lots of useful features like that.

However from my experience with Oracle and from what I've heard about their source code, its fundamentals are just disgusting. Decades of haphazardly piled-up features and fixes means that if you're the lucky customer to find a new bug, good luck understanding what the heck is going on by yourself.

I'd like to think MSSQL is in a bit better shape, but they have lots of the same incentives so who knows.

Oracle is like America. Its not all the same. The database itself is now one of the most rock solid pieces of software ever built. Other products - yeah, less so.

I think this is perhaps referencing a recent HN comment about the code quality of Oracle DB: https://news.ycombinator.com/item?id=18442941

Ha, that's pretty interesting. Not too sure how much credit to give it though based on this:

> It takes 6 months to a year (sometimes two years!) to develop a single small feature (say something like adding a new mode of authentication like support for AD authentication).

There's no way in the world I would have imagined that adding AD authentication to any database would be a "single small feature".

For a long time Postgres was the quintessential example of how open sources projects are driven by developers who love to add all kinds of cool, whiz-bang features... but there wasn't boring enterprise features like incremental backups because that's no fun to write and test. But that characterization is long out of date. I think.

plsql, packages and 'as of timestamp'

> What features of Oracle or SQL Server would you say it is lacking?

An ass to drag onto the carpet. In enterprise speak, "support" means "legal liability". You have to be able to sue someone if it fails and loses your company money. This was taught to me by an old IT hand back when I was a young naïve Linux kiddie "you have to be able todrag somebody's ass onto the carpet when the shit breaks".

Oh yes, and there's also brand recognition and a large pool of certified Oracle or SQL Server DBAs to draw from when making hiring decisions. Yes, technical people will know that Postgres is solid and will know how to spot good people to maintain a Postgres installation -- but the person signing off on the IT budget is a total normie. He will smile when he sees familiar names and make a lolwut face when he sees postgres.

Relatedly, Postgres doesn't have a yacht. Oracle's yacht won the America's Cup.

I’ve worked for dozens of enterprise companies. Support means support.

It means that anyone in the company can ring a vendor 24/7 and get the best quality help on the product. It means your outages are smaller and less frequent. You usually can’t get that with a generic open source product or from smaller companies and it’s especially an issue issue in non-US countries where it’s often impossible to get decent support.

I’ve never once seen a company sue a vendor or vice versa.

> I’ve never once seen a company sue a vendor or vice versa.

The first three results I got from a search engine:

1. https://palisadecompliance.com/oracle-lawsuit-pushing-cloud/

2. https://www.theregister.co.uk/2018/03/22/oracle_shoddy_servi...

3. https://www.businessinsider.in/I-felt-like-we-were-being-ext...

Including audits with threats to pay ever increasing amounts every year?

Ouch. Also add in that the budget changes dramatically in year two...

Every enterprise product I’ve ever seen has different prices for each year.

And if you can’t handle this I think you have bigger problems to worry about.

>And if you can’t handle this

What a retarded position to hold.

You have forgotten what the purpose of these providers was in the first place.

"Company X provides service/software Y so you can do business"

But that's not what they do.

"Company X optimizes to extract as much rent as possible from its client in order to show ever increasing amounts of profitability to the stock market"

You do know that most vendors offer increasing discounts in later years of the contract. It’s not just that prices increase.

And I’m guessing you’ve never dealt with a vendor before because often companies negotiate better terms than list price.

Of course, when managing my product or running my business, chasing discounts for and (re)negotiating with the vendor of one of the core components of my product/service are things I very much look forward to.

My mistress is retarded and I find this comment very offensive.

Why is flagging the fact that the term retarded is offensive to some people getting downvoted?

Maybe I've just seen the inside of too many "enterprise" companies. But I suspect the most important enterprise feature is "Supports the architecture that we came up with 20 years ago and that was the hot choice when the CIO first started coding."

Buying Oracle or SQL server is often more about compatibility with other products or specific features.

And CIO/CTOs rarely unilaterally make decisions on technology choices. You have Enterprise Architects for that.

Sure, but I think it's a CIO's job to push an organization forward in technology. But I have seen plenty of places whose current reason for using Oracle is, "Welp, that's what was hot when we got started and now it's baked in to our architecture." And that's the same reason I've seen older "enterprise" companies keep their mainframe stuff going.

In contrast, I recently heard about a company with a market cap in the tens of billions where the execs said, "By date X, we won't own hardware or run datacenters anymore." I'm told it forced a lot of teams to reevaluate fundamental technology decisions made decades ago.

So sure, the decisions are proximately about compatibility or specific features. But I think it's reasonable to ask why those specific features are necessary. Because places like Google and Facebook make it clear that it's not scale alone that makes it necessary to give Oracle a truck of cash every year.

Reply on news.ycombinator.com