UrbanCode Deploy Reviews

The ability to deploy as often as possible allows a number of people to work on it at the same time.

It's provided us with more reliable and faster deployments, as well as the ability and flexibility to create and modify deployment applications to meet the needs of the ping.

We're running version 4.8.4, which isn't the latest version. This version lacks reporting, which may be improved in the latest version.

I've used it for four years.

It's been a very stable product. We've had a few performance issues, but they haven't been serious. I'd say we've been going for a couple of years without having any issues or needing any support or assistance.

It's absolutely scalable. We have gone from about 17 products to 32, maybe it was 42 since we've gotten uDeploy. Before it could take a couple of days and we could only do a few deployments, just barely a few in a month. Now we can do multiple deployments in a day, we can get hotfixes out very quickly, and we can easily modify and customize processes as the needs of the teams change. It scales very well for us. We also, for instance, have needed to improve our security and all that is there in uDeploy. We don't have to do anything other than figure out how to configure it.

Technical support is always very knowledgeable and proactive. In contacting different people, we've always had very good responses.

We weren't using anything before and we had to get something up and running quickly. It did take us a little longer to get the uDeploy up and running than we'd hoped, but once we did it certainly met all of our needs.

It was complex at times. It took us a while to come up with the standard process that we needed, but then once we had it it also took a while to actually get all of our projects up and running. We didn't know how to use the CLI at that time. We might have been able to do better if we knew that.

It was all done in-house. I'll add a comment that the learning curve for uDeploy is high. Even now there's a couple of us who know it well, but being able to bring somebody else on-board and explain it to them would take a little time. Of course, now that we know what we're doing, we have somebody so we can explain it to somebody else, but at the time when we started and since we didn't know what we were doing, and it took a lot of trial and error to figure out how to use it.

Standardize processes as much as possible so that you can re-use them. Don't be afraid to create multiple applications when that reduces the complexity. Don't make applications really complex to try and do all these things all at once. If you can, break it down into smaller apps because they're easier to manage. Generally the more independent things are then the fewer things break at any one time.

If you have problems, it doesn't affect more than just the particular instance that you're looking at. Our developers have done a lot to break down our projects into smaller and smaller pieces so that they help with that. We can deploy an application, but if it's only got three modules that are changed, then we only need to deploy three modules. Breaking it down into independent pieces and reusing them, I think, is really important.

Integrating it with an automated build system, I highly recommend that. We use Jenkins. We can easily connect our Jenkins artifacts and push them to uDeploy so that they can be deployed. We have some interaction between the two so that we have good auditing and knowing which build is the one that got deployed. In fact, using versions that relate to the build version, so that when you go into uDeploy and you look at a version then you can see that that version came from specific build and the build tool, that's really helpful.

It just changed my whole job and made it easier. I'm sure that some people might have different opinions on that, but coming from where we were and getting to where we are was just never could have been done without uDeploy and it's just incredibly powerful for us.

• The gating and approval workflows.
• The udclient.jar API calls are relatively easy to understand.

• Centralization of our deployment process
• Unification of our nonprod and production deployment process

It makes it easier to create new resources, especially new components, and to import new servers into uDeploy Environments.

I have used it for six months.

Once I understood the API, deployment was reliable.

We have not encountered any stability issues.

We have not encountered any scalability issues.

Customer service is good.

N/A: I have not had to invoke technical support.

We did not previously use a different solution.

I was not involved in the initial setup & configuration.

We use IGS on site to assist our IT staff as needed.

We are still evaluating competitors like XebiaLabs Release, Puppet and Chef Automate. There is a chance we could move to Chef Automate

Learn the API and automate common tasks from your CI|CD pipeline

We like a few things, such as the calendar, the permissions model, and the process editing. The component template and deployment interface are very good.

Our organization has already improved by quite a bit by taking out so much manual intervention from multiple people doing the deployments. We've simplified our process and the deployments are now much faster.

Sometimes the deployment process reports are faulty. I've run into situations where the job says it was successful but it really erred-out. It would be nice if the reports are correct all the time.

The configurations tab takes forever to load. It doesn't work very well so that needs to be improved. Also, you can't filter by resource tag, which would be useful.

We've used it for six months.

The deployment reports are faulty.

So far, the product has been very stable.

We're in the process of moving pretty much most of our codes to run through it one way or another. I anticipate it will scale well.

So far, we haven't had an issues with support so it seems okay.

Our existing solution was mostly manual deployments so that, of course, is very error-prone. We looked at several different solutions to try to see what we could implement quickly and what would have the most impact, the most return on investment in a relatively quick amount of time. This was the tool we decided to try out and so far it's working pretty good.

It's very time consuming, the setup of the solution. We ended up scripting a lot of it, so that it was quicker. There's a lot of pieces that you have to set up for one particular job to run on it. You can actually make steps and end up having to sit there and figure out what you missed, so we ended up scripting a lot of it so that we wouldn't miss pieces, which made it also quicker to set up new applications, to be able to get all the way through deployment.

We implemented it with our in-house team.

It's a pretty straightforward tool for straightforward deployments. If they have many complicated deployments, they're going to really want to look at the plugins and what's out of the box to make sure that it's going to be able to do what they need to do, and that they need to make sure that they have the resources and time to spend setting it up.

There are three main valuable features for us:

1. The ability to deploy objects consistently to any of our multiple environments;
2. Prior deployments and logging stored in a web GUI which can be accessed by any member of the team; and
3. Flexibility of the plugin system to use features supplied by IBM or the ability to write our own.

Teams are now able to schedule deployments to any environment and watch in a screen the deployment occur. This has allowed teams to reduce the number of staff online when a deployment occurs.

Tracking and audit-ability of the deployments are easy and allow teams to research any issues quickly and resolve the issue with little assistance.

The reporting functionality is limited and it is difficult to retrieve information from the database due to the table structure.

We've used it for about one year.

We encountered no issues with deployment.

We have occasional minor issues with stability.

We've had no issues scaling it.

6/10

7/10

The prior solution was created by our company. We determined that to add the features required to expand our solution, it was better to use a vendor product.

The initial setup was straightforward and well-documented.

We implemented it through our in-house team.

I'm not able to answer this.

Cost is based on the number of agents installed. This is normally one agent per target box/system.

Yes, we evaluated Jenkins.

This product is excellent for deploying components and running processes related to deployments. It is not intended to be a build system, but is easy enough to incorporate into any CI Build system. The tool has two "sister" products which work well with UrbanCode Deploy. It is intended for this tool to be used for larger companies.

Getting the deployments automated adds a lot of value. Before we automated deployments, there was a big problem with manual intervention and people making changes on production services and forgetting what they did. It was just impossible to reproduce deployments.

Then there's the traceability aspect -- you can trace back to the code which is stored in ClearCase through the build. All the deployment artifacts can be traced back to the original source.

The biggest area of improvement that I would suggest is in the upgrade process. We started 2015 with uDeploy 4.7, and in order to get the latest version, we had to upgrade first to version 5, which was introduced for the IBM branding. That was straightforward enough, but the jump from version 5 to version 6.1.2 was enormous. We had to focus on testing the security model because it was completely new. We documented and tested it in a production environment. Once we did the upgrade, we found another big change in the version implemented, which we didn't anticipate and which caused problems in our production environment. We re-architected the implementation and there were 4-5 failings we had to work through. Future upgrades probably won't be this difficult, but this one was tough.

In terms of functionality, the reporting could be improved a bit. It seems like the new version has better reporting capability, but data visualization would be nice to have.

We first started using uDeploy around 2011 or 2012, but when we started to automate our deployments in 2010, we used Anthill Pro. When we started using uDeploy, we kept Anthill Pro as a backend for the builds. We now are on version 6.1.2 of uDeploy.

It's less stable than our other IBM products. We have ClearCase, and it rarely is unstable as it's designed for devops work and can deal with the development process on the product itself. There have been a lot of broken stoppages in uDeploy 6.1.2, particularly in the upgrade. We installed the newer 6.2.01 in our sister environment and actually had some regressions in the interface. It seems to have the stability level of open-source products.

It's scaled fine for what we need it to do. We have over a hundred applications configured in it.

They're always knowledgeable, but it seems they're overwhelmed. They're great and deal with our issues, but it seems the hurdle is their other customers who also need help. There aren't enough resources available.

It's really good if you have a lot of applications that need to be deployed in different environments. You need to have a team to support it because there's a lot of different pieces and it's a really deep solution. There are many features, only some of which we've been using. It's not the kind of thing that only one person can work on. It's not for a small shop; it's for an enterprise.

Plugin-based architecture.

It has provided a platform for traceability across the whole release cycle for more than 700 diverse applications.

• Licensing structure - pay per machine deployed to.
• Additional integrations with ALM solutions.
• Unclear why it doesn't support build automation, which is covered by a separate product.

3 months.

No issues encountered.

No issues encountered.

No - scalability is handled via client relays which offload the server's work to proxies.

Good.

Good so far. Not a lot of history there.

Yes, we used Build Forge. It's a build tool that was being re-purposed for deploy and it was beginning to fail often.

Very straightforward. Command line prompts to install all components, and hook into enterpriseDB. Trial version will use its own Derby DB.

TBD. Have handled most of this in-house so far, but we have a service contract available. No info on expertise.

No specific numbers right now, but we're saving the deploy team, the infrastructure team, the help desk team, and the development teams many hours. Results are variable based on how much automation was already done, and the quality of the legacy code.

Evaluated Automic, CA product offerings. Also kept tabs on the Serena offering.

• Start with small wins to build momentum towards the larger applications.
• Align with the compliance team from the beginning. We had compliance and internal audit in the room for the PoC to run through concerns up front.

• Rich plugins
• Visualized process design
• Visualized approval process

It has automated the deployment of tasks to different environments which we were previously only able to do manually.

They need to reduce the footprint and improve the performance of UD agent. If the agent runs for too long it can cause a memory issue on the production server. We must keep the agent offline and only enable them during deployment.

We've used it for three years.

There have been no issues with the deployment.

There was no issues with the stability of UD agent.

We have had no issues scaling it for our needs.

In my experience, I'd say that technical support has been good so far.

The initial setup is easy and straightforward.

We implemented it with our in-house team.

We compared UrbanCode Deploy with XebiaLabs XL. We chose UCD because of the process design function.

The product became more complex after IBM acquired it.

The valuable features are “scriptability” and customizing the deployment processes.

It’s not necessarily the product, but more the drive to automate deployment that results in improvements.

UCD gives the freedom to create reusable component templates. You set up a process for deploying something once, such as a standalone Java application, and then that is “templetized” and can be reused.

In these templates, you can:

• Standardize the used repositories, such as Nexus
• Standardize how an install.sh script must be made
• Determine how staging parameters must be stored in a config file.

Many of the improvements are, therefore, based around automating the deployments. They are automated in such a way that no more "screwdrivers under the hood" are allowed in any stage.

This saves time, and makes the process much more reliable, reversible, repeatable, and traceable. In the beginning, this is painful. I can’t stress enough how much effort should go into getting this right.

WebSphere deployments, for some reason, don’t work out-of-the-box. We have worked on the Websphere issue with the IBM uDeploy development team a lot more now. What we want to be able to do is apply configurations to the Websphere using the standard plugin. This can be done by creating json snippets that must be parsed with the large json files of the cell, the node and the server that have been created during the mandatory initial configuration discovery of the target machine. We have had lots of difficulties getting the parsing to work, now a new version of the config plugin has been released which is an improvement. 

However, what we want to be able to do with our CICD automation is to create configurations paired with the EAR files so that we can start doing partial updates, of only the parts that have changed. Also rollbacks will this way be much easier to accomplish. uDeploy can not work like this to date, the plugins do not allow it.

UCD needs to perform a discovery of the environment. This would not be needed if it would understand more about WebSphere environments and releases.

We have used this solution for one year.

In terms of deployment for WebSphere, the configure plugin didn’t do what we wanted. The plugin requires a discovery of the target WebSphere environment. For some reason, applying changed configurations via the plugin doesn’t work for us.

In itself, UCD is a stable tool. Once something works, it continues to work.

I would give customer support a rating of 6/10. The customer is expected to bring a significant amount of knowledge to be able to configure component templates, resource tree, etc.

Customer support is available, but it is remote and only acts upon raised incidents. At our own cost, we have hired IBM specialists on premise to solve the WebSphere issue.

We didn’t have a previous solution. This was our first real attempt to introduce one central deployment tool to automate and standardize deployment processes for all techniques, such as Linux, IIB, IIS, and WebSphere.

We chose the product because we have a long-lasting relationship with IBM.

Initial setup was done on only one environment. Even then, UCD has a quite complex setup due to a needed high-available setup with load balancers, queue managers, license servers, and databases.

Depending on the size and complexity of the organization, you need at least three environments:

• To develop and test new versions of UCD
• To build reusable deployment solutions
• To execute them

IBM did the implementation. Unfortunately, they did it without considering that deployment automation is not just about a tool, but much more about standardizing and optimizing the deployment landscape and the processes.

It was done as a remote implementation, which of course didn’t fit. It had to be changed in numerous ways.

I have no knowledge on the ROI. In the end, I think the costs must be seen in the light of the objective you want to achieve. If you’re considering release management, CICD processes, and want to be a DevOps organization, then the costs for the tool don’t matter much.

Think about what deployment automation really is. It means no tweaking throughout the stages whilst applying changes. Everything must be code. That is the most important step; having everything as code.

Once that is done, then probably all of the good deployment tools in the upper-right corner can do the job.

In the end, deployment should be something that runs in the background; getting a signal to deploy something that has been created.