We recently adopted Librato as a monitoring solution, and while the interface isn't quite as flexible as DataDog, the pricing model is incredible: you pay only for the metrics and frequency you need. https://www.librato.com/pricing
For instance, we only need low-frequency sanity CPU/memory checks on hosts running stable microservices, which we can do for ~$1/host/mo, but for our heavily-utilized core databases and web servers, we have a lot more, including full Mongo and Elasticsearch metrics supported out of the box. Total pricing is transparent and estimated for the month on what seems to be only a 24 hour lag. Combined with https://logdna.com/ (YC W15) for no-nonsense, beautifully presented log aggregation/search/introspection (which itself is pay-as-you-go per GB ingested), we can pretty easily see what's going on, with completely elastic and reasonable pricing.
At the end of the day, there are only so many ways to put charts and logs on a screen. There will be room for value-add services, say, using machine learning to see warning signs even if not explicitly set up by the tenant. But for those doing basic monitoring and alerting, I think the space will increasingly become commoditized the same way IaaS has been... and at that point, people will insist on transparent pricing that corresponds to some predictable premium on the actual burden they're placing on their provider. And charging by host will be only barely correlated with that. But until that time comes, you can't blame DataDog for being aggressive. It costs money (and, occasionally, the blood of smaller users) to be winner-takes-all, after all.
We're datadog customers and we still get calls and emails from their sales people asking us what monitoring tools we use and if we're interesting in switching to datadog.
They must not be using any sort of CRM, or their salespeople might not do any homework before cold calling.
One of their sales people tried to social engineer contact with me when I was running ops at a previous gig. They called out customer support line and said they were "just disconnected from the ops guy" or something of the sort, and asked to be transfered.
That's possibly one of the stupidest things I can imagine to sell SaaS to security-minded people, and they were surprised and offended when I told then to fuck right off.
Please don't do business with such unethical people.
For instance, we only need low-frequency sanity CPU/memory checks on hosts running stable microservices, which we can do for ~$1/host/mo, but for our heavily-utilized core databases and web servers, we have a lot more, including full Mongo and Elasticsearch metrics supported out of the box. Total pricing is transparent and estimated for the month on what seems to be only a 24 hour lag. Combined with https://logdna.com/ (YC W15) for no-nonsense, beautifully presented log aggregation/search/introspection (which itself is pay-as-you-go per GB ingested), we can pretty easily see what's going on, with completely elastic and reasonable pricing.
At the end of the day, there are only so many ways to put charts and logs on a screen. There will be room for value-add services, say, using machine learning to see warning signs even if not explicitly set up by the tenant. But for those doing basic monitoring and alerting, I think the space will increasingly become commoditized the same way IaaS has been... and at that point, people will insist on transparent pricing that corresponds to some predictable premium on the actual burden they're placing on their provider. And charging by host will be only barely correlated with that. But until that time comes, you can't blame DataDog for being aggressive. It costs money (and, occasionally, the blood of smaller users) to be winner-takes-all, after all.