An Open Letter to Modulus.io

deployment
1

Dear Modulus.io,

We don’t understand why you aren’t doing more to actively capture the market for Meteor hosting. Meteor has taken off and there are not many great hosting options. You guys are really missing an opportunity to make a name for yourself as the leading Meteor host. You wrote a couple articles about Meteor two years ago, and then nothing since then (including failing to update those articles regarding changes to the demeteorizer command line). You haven’t even done any basic SEO/SEM like putting “Meteor hosting” on your home page (until I wrote your CEO about it a few months ago). You could also publish many articles on Meteor to boost SEO, and actively engage the Meteor community so you can better understand their needs. Take a stance and boldly explain what makes Modulus a great Meteor hosting option.

A lot of people seem to think Heroku is the best option without really knowing why it isn’t. Seems like an opportunity here. Run a “Switch to Modulus” campaign and give people a month or two discount for giving Modulus a try. Or you could offer something like: “We’re so confident with our Meteor hosting, that if you migration your app from Heroku and decide you don’t like us after 30 days, we’ll give you your money-back.”

http://screencast.com/t/6UlechCm9tT

If Modulus was my company, I’d have made the strategic decision to be the number one host for Meteor apps starting a couple years ago. Early on we recognized the revolutionary nature of Meteor and have been doing all new development in Meteor since version 0.6. We have several Meteor apps which have been in production for 2​.5​ years already. We don’t understand why Modulus has taken such a passive role when it comes to seizing this opportunity. Meteor is barely mentioned in your blog and doesn’t show in the your tag cloud:

http://screencast.com/t/o8bJtiBbxS3z

This seems like a major oversight to us.

​Why are there no Modulus reps talking back on the very popular Meteor forums where Modulus is heavily discussed:
https://forums.meteor.com/search?q=modulus
​
As evidenced from the design of that last email survey​ (“Fill out this 5-minute Survey for a $5 Gift Card!”)​, it seems you guys don’t even want to really know what your customers have to say. There was no opportunity within the survey to provide any feedback about your service. I can only guess that maybe there is a culture at Modulus to reduce interaction with customers (due to the high costs of customer support?) I can appreciate that, but maybe there should be exceptions for marketing and strategy teams​ right?

It took Modulus 1.5 years to reply and they still didn’t answer the question :wink:

​Did something really bad happen there after Progress took over? Did all the innovative employees leave or get moved to other departments?​ I’d really like to see Modulus take the Meteor hosting opportunity seriously especially in the area of better application performance monitoring for Meteor apps. Let us all know what we can do to help. Meteor application hosting is an underserved market.

Cheers,
Max
Founder, White Rabbit Japan [Built with Meteor, hosted on Modulus.io]

UPDATE: I see Modulus added “METEOR SUPPORT” to their homepage. It wasn’t there when I originally wrote this letter to their CEO a few month ago. So maybe my letter did have some impact, although no one replied to it.

1035×252 48.2 KB

6 Likes
2

Because Galaxy is coming out

2 Likes
3

Modulus started hosting Meteor over two years ago. Galaxy still doesn’t have a commercial product, and you have no idea what the pricing will be like when they do. Galaxy is described widely as “an enterprise-grade Meteor hosting platform.” I interpret that to mean: expect Enterprise-grade pricing.

Meanwhile Modulus.io has had over two years to put miles between them and Galaxy. It’s nearly a year since Meteor hit version 1.0 and still no Galaxy.

1302×814 578 KB

2 Likes
4

I agree with @maxhodges, I tried to use Modulus late last year, and found the docs out of date and confusing. The platform seemed only to semi-support-meteor deployments and after all that the cost was off-putting.

I ended up using AWS+Ubuntu+MUP and have had no issues so far.

I don’t know much about Galaxy (I’m not sure many do), but if locked into a specific cloud, and have pay for the tool, I’ll probably just continue using AWS.

5

Also I’ve found them to have the worst support for a commercial saas product I’ve ever seen. So, i think they’re solely focused on being the low cost provider whereas heroku wants to deliver a bit more.

They are really really easy to get going with right up until you have a problem.

1 Like
6

Exactly my experience. I tried to call them and they told me I’m basically on my own with the exception of the forums.

On the other hand with AWS, when I ran into issues on start, I gladly paid $50 for a month of on-demand full support and got all my issues resolved. They they had competent support staff and stayed with me through the ups and downs. In the end I cancel the support half way through the month and the next month when I had another issue, they gave me credit for the prior month. Now I’m a loyal customer and if all goes well I’ll spend a lot of money on cloud services with them.

Great service goes a long way.

1 Like
7

That’s a bummer to hear (about Modulus). Hopefully Galaxy turns out great and is reasonably priced. I’ve got a major web service I’m working on now, and I’ll need a stable platform for it. If Galaxy is outside my budget, I’ll probably just spin up a Vultr VPS and use mup/mupx to deploy, but host the database on MongoLab. (normally I may have gone with Compose.io, but their lowest price point is moving up to $31/mo which is too high IMO).

8

Hey All,

I’m one of the founders of Modulus and original author of demeteorizer. I just wanted to jump in and let everyone know we are following this thread and taking everything you say very seriously. We are actively developing a much better Meteor deployment model that I think you guys will really like. It’s based on our new build/run image system we launched a couple of months ago and open-sourced here: https://github.com/onmodulus/docker-images

It stinks that some of you had bad support experiences. Joining Progress was very good for us and we’ve grown rapidly. Maintaining proper support quality ended up being our biggest challenge as we’ve grown, but definitely something we are working on.

Again, thanks for the awesome feedback and I’m really happy to see the Meteor communities cares enough to spend the time and energy letting us know how to provide a better product!

12 Likes
9

We found deploying on Modulus fairly easy. Heroku took me more time to get started. But yeah I was a little confused too after they made an undocumented change to the CLI. If you ever want to try deploying to Modulus let me know and I’ll help with any rough spots. Once it’s setup we just deploy new versions with a few keystrokes

type modulus deploy

maxMBP:wreClient maxhodges$ modulus deploy

select a project

Welcome to Modulus
You are logged in as max@whiterabbitjapan.com
Please choose which project to use:
...
6)  WRE-PRODUCTION-ADMIN                   Node.js
7)  WRE-PRODUCTION-CUSTOMER                Node.js
8)  WRE-TEST-ADMIN                         Node.js
9)  WRE-TEST-CUSTOMER                      Node.js
[?] project 7
Selecting WRE-PRODUCTION-CUSTOMER

. . . wait for it to build, provision and deploy . . .

[✓] WRE-PRODUCTION-CUSTOMER running at xxx-49425.onmodulus.net
10

I’ll chime in and say that I’ve also had very good experienceswith Modulus. And @vibledux, thank you very much for all the great work with demeteorizer and the docker image infrastructure.

While we have your attention, I understand from one of my clients that you all are willing to sign HIPPA BAA agreements. Are you able to confirm that? Will any HIPAA specific offerings be available with the new docker image infrastructure?

If so, there’s a growing pool of healthcare projects in the Meteor community which are leaving prototype phase, and preparing to go into production. Galaxy apparently doesn’t have plans for HIPAA support, other than the enterprise licenses to deploy behind-a-firewall.

So those startups looking for HIPAA compliant Meteor hosting are evaluating between Modulus.io and Catalyze.io. I’d love to be able to tell those startups that there is an alternative scale-out path available for them, and they’re not locked into Catalyze.io.

3 Likes
11

I’ve had a decent experience with Modulus as well. I think it took me about 5 minutes to get an app up and running. The only problem I had was with Mongo crashing… SO I switched to compose for hosting the database and everything has been great since! The app I have hosted with Modulus is pretty big and I haven’t run into much trouble.

1 Like
12

We absolutely do sign BAAs. We have HIPAA customers in production now and I do believe at least one of them is using Meteor. I agree that Meteor appears to be picking up momentum in this area. Our HIPAA customers typically live on completely dedicated installs of Modulus on AWS’s dedicated infrastructure. We have a BAA in place with Amazon. You can also do a totally on-prem version of Modulus if you have your own datacenter.

4 Likes
13

I’ve had excellent and terrible experiences with Modulus. The latter led to me transitioning my entire infrastructure to Digital Ocean over the weekend.

tldr; Do not use Modulus for serious products. When you use PAAS, you’re screwed if something goes wrong (which invariably it will with something like fast-changing Meteor) and the support isn’t there (which, right now, it is definitely not).

In 2013, Modulus was a young company that was working hard to entice Meteor developers (they deserve kudos and gratitude for the open source projects). I took the bait, and ended up running a few toy apps on their infrastructure. I dealt (I think) with the founders of the company (including @vibledux) who were blindingly fantastic. On one occasion, my app was inexplicably timing out, and one of them must have spent 5 or 6 hours digging through the system over the weekend to find the issue. It was something that other people were likely to experience, and thus important for their product generally, but I was still super impressed with the commitment.

Fast forward to 2015, and I am running a fast growing app with paying customers. Modulus started to have intermittent issues early last week. Serious issues. I spent 4 days with paying customers unable to get onto my app intermittently.

These were the problems:

  1. That much downtime is totally unacceptable for a PAAS provider (to be clear: when I tried to go to their web support channel on one occasion, that was down too);
  2. Responses to queries were super slow and unhelpful in a critical situation - maybe my issues were worse than everyone else’s on account of my particular code base, but it was impractical to find out with the response times; and
  3. Public status updates on the issues were garbled at best, and dishonest at worst.

Herein lies the rub: I’d like to be able to fix the issue myself, but when you use PAAS, you’re not able to. You are completely reliant on the people running the platform. When they don’t perform, you’re screwed.

Realising that things were not going to get better, I transitioned everything to Digital Ocean with MUP. Which is bloody fantastic by the way. A beautiful place to be between complete configurability and total ease of setup. Everything is running perfectly again, and I no longer feel like I’m playing roulette when I hit my site.

To pre-empt the usual tech company response: no apologies and overt empathising, please. As founders of your company, I feel bad for you. You’ve handed over to a larger company that doesn’t give a shit about the reputation of what you’ve built (which was great).

Now that your overlords have obviously de-prioritised keeping support levels up, all I want to hear about is when you have guaranteed response times for support when people are willing to pay for it. It’s the only way I’ll use the platform for anything serious again in the future.

1 Like
14

HI @vibledux,

I’ve been using Modulus for several minor apps for the past couple years, but a few weeks ago we deployed a mission-critical Meteor app and the performance is often sluggish and unresponsive. I can’t understand why this is happening. According to the Modulus metrics, why CPU never goes over about 2% and RAM usage is well below the max.

I signed up for Kadira in order to try and get deeper insights, but nothing is really out of spec there either. CPU, RAM and response times all look very good, but even when the actual experience of the site is very poor. The database is hosted on MongoLab and nothing there would indicate any bottlenecks. It’s a complete mystery. We can only imagine it’s some kind of hard disk contention, network congestion or server availability . . . but whatever it is it is not showing up on my analytics so we’re helpless to address it.

We upsized our servo on Friday just to see if it would help, even though clearly the CPU and RAM are far from being maxed out. But unfortunately it didn’t seem to help.

We also changed from Joylent to AWS just to see if that would help, but we haven’t noticed a difference.

I guess this ‘premier public hosting’ service is some kind of shared server. So are we basically at the whims of whatever our neighbors are doing on the server? I wrote to get pricing for the ‘private hosting platform’ but the reply I got was,

“What are your questions specifically in regards to pricing?”

Well for starters: how much? (duh!)

It seems pricing for private hosting is too sacred to be communicated by email, so I have to wait until we can schedule a telephone call between Eastern Standard Time and Japan Time.

After reading the testimony of @pjbrow I’m thinking perhaps I should also give MUP + Digital Ocean a try. If you want to keep us as a customer let me know if there is anything you can do to try and help us out. Maybe by working together you can learn a bit more about where the potential bottlenecks and gotchas are with Meteor app performance.

But I understand if you can’t help. Hosting companies need to draw some kind of line between providing infrastructure and supporting and troubleshooting application specifics. However if something runs better on MUP + Digital Ocean than Modulus, I gotta make the best decision for my company.

731×852 475 KB

15

@vibledux
a few hours later . . .

Here you can see my server just flatlined

704×861 466 KB

looks like this in Kadira

1322×778 115 KB

in the browser it simply says

Unable to connect to any application instances.

I clicked Restart in my Modulus admin, but that is stuck too

728×79 34.9 KB

What am I suppose to do?

16

Please submit a ticket to help@modulus.io with the details of your project so we can investigate.

19

I have been using Modulus for most of 2015 and would echo the other users here in saying that the support is pretty horrible. The only reason I’ve stayed on is that there aren’t any great alternatives. That will change when Galaxy comes out.

With that said, I actually find your PaaS service very useful. It’s just really rough around the edges and when problems arise, I don’t get a reply from support for a few days to a week. This is unacceptable for real businesses that are reliant on reliable uptime. To be honest, if you guys had support that consistently replied for critical issues within an hour and noncritical issues within one day, I’d probably stick with Modulus even when Galaxy goes live assuming the price is comparable. As it stands, I continue to run into problems that block my development and where my users can’t access the service without anyone to help answer my questions.

I can guarantee you that the incremental cost of hiring a few more support engineers will be marginal compared to the amount of existing and potential customers you will lose by not making a commitment to improving your overall support efforts.

1 Like
20

I wasn’t able say this before because I was under NDA with MDG about Galaxy. We got early access to Galaxy, pricing is out of this orbit.

1 Like
21

@maxhodges wow, this is out-of-reach-amazingly expensive for a small startup.

23

Galaxy finally announced pricing for a cheaper Developer plan.

Developer Edition includes many of the features our production customers love, including:

  • One-command deployment
  • Push-button container scaling
  • Coordinated version updates
  • SSL capabilities
  • Basic application metrics and logging
  • Basic support

Developer Edition will be priced on a pay-as-you-go basis of $0.035/GB/hour (about $13 per 512MB container per month), initially limited to a maximum 5 x 512 MB containers (spot instances).

[Invitation here]((http://info.meteor.com/lp/galaxy/developer-edition)

Footnote: After migrating to Galaxy, we no longer run any apps on Modulus.io

1 Like