First impression and feedback

Hi All,

I migrated my home box from MR7 to v17 and all good.

In my case, the IPS did not start automatically after the reboot.

The other thing is the UI is faster than v16 but the Network menu is very slow compared to the others. I have only 2 NICs and 2 VLAN.

Last thing, where is the policy test?

Parents
  • Hi All,

    in my personal opinion i will change:

    - NAT and WAF Rules on different pages. Firewall ACL is another security concept

    - on Dashboard no Health status about Power redundancy

    - on Dashboard no Health status about disk, or mirror disk. Only one way to discover problems is to go on DataCenter room and listen to the alarm...

    - on Dashboard on Web Hits, we would see the number of HTTPS connections and HTTP

    - We need a full log export, in case on Deep analysis on Forensic analysis. Reports are goods for Executive and for POC but you partner need to be able to answer who did what,when, wich protocol and wich port where used.

    - IPS Engine/Policy. If you need to exclude a single signatures only for a restricted number of users/pc you need to create two rules and play with priority: ok, but if you are on middle market customer how many rule you need to do to secure the customer? This is the same for Application policy

    - About metrics, decide to Use Kbit (kb)or KByte (KB), but with the right Sintax.....If you play whit BWM too many misunderstanding on the GUI and Documentation

    - Help us with O365 creating an Hidden Feed RSS to automate the download and the population of O365 IP/FQDN service to exclude from Proxy. Too many errors and problems about it.

     

    Thanks

  • I have always complained about the static gui. It shows very limited information and most of it is not important as a snapshot. Also completely agree with kilobit and kilobyte problem. It is really not that difficult... ALL live traffic including QoS rules should be in kilobit/mbit since we get the traffic from our ISP in kilobits/megabits and our network cards are also in megabits/gigabits etc. However the aggreagate traffic, like how much traffic did TOM use or the total amount of traffic should always be in kilobytes/megabytes etc. Maybe v18...

     

    Edit: On a side note, I have a different bug report about the firewall passing all traffic as soon as it is connected to the network. Did you guys know about this behavior? It has been there since v16[:#] https://community.sophos.com/products/xg-firewall/sophos-xg-beta-programs/sfos-v170-beta/f/sfos-v170-beta-issues-bugs/96108/bug-firewall-starts-passing-all-traffic-before-running-the-wizard 

  • Luk, the basic OS is ubuntu (atleast the kernel part). Its the daemons that they want to reinvent and keep carrying from cyberoam. I guess for compatability with cyberoam CLI. For example VPN, if they had chosen openvpn open source code, they would already know what are the limitations and they could have tailored it to their own requirement with the huge number of programmers they have on their payrol. But they reinvent everything or atleast try to improve the functionality of existing daemons. I guess they think they will have to do it eventually (like http proxy daemon in UTM) so might as well start now. Problem is that writing a daemon is a lot harder than writing shell scripts and its showing up in the speed of development of XG.

    MTA is the prime example. They took awarrenMTA, which wasn't designed as a mail server daemon in cyberoam and have been trying to fix it for two years. In the last year, they added the functionality of smarthost. I just shake my head... there are so many open source MTA servers available, heck the UTM has an open source MTA that has been rock solid forever. Recompile it or postfix for XG and then add cli integration later. Most of the cli stuff is not needed anyway if gui was capable of handling this stuff. Cli is nice in cisco routers or even vyatta when it was alive but XG strength is its gui. Vyatta was successful because they had juniper cli and still linux underneath. XG has limited cli and they have completely locked the linux part down. Someone was trying to change the NDR times on XG, not a problem in UTM or any linux. In XG... I don't think so. What good is CLI if it hinders instead of helps with the OS. I hate CLI anyway. I like gui, and there are billions of people who also like gui otherwise apple and samsung would make fancy CLI instead of beautiful gui. I don't want to look at packet captures etc. I want the gui to tell me that the packet got dropped for such and such reason like the UTM does. Is it too much to ask? Packet capture on a client... yes. Packet capture on the firewall because the logs don't tell you where the packet went is incompetence[:'(]

    Another problem I have is that since the daemons are created in house, they keep giving friendly error messages like here https://community.sophos.com/products/xg-firewall/sophos-xg-beta-programs/sfos-v170-beta/f/sfos-v170-beta-issues-bugs/95959/bug-404-page-error-translation same with MTA and other daemons. Lets keep the errors to industry standard and not try to make them easier to read. I was reading a thread about MTA where the error was the remote server abruptly ended connection. If they had used error codes, it would have been easy to troubleshoot but now this could mean anything. That is why they are having a hard time producing UTM style logs. If regular error codes were being generated, grepping the log file is easy but not if you are making your own error names.

    I am sure sophos is aware of all this and every beta we complain about the same items. The web portion of the team that  is a part of has always done a great job. Its the low level kernel tweakers and daemon writers that can't produce stuff fast enough so that the people that work on gui can impelement that stuff on top. 

    They need object based items like UTM started in v7/8. That made UTM suck for a little while with memory leaks and slow downs in conf daemon but it was necessary. You can name anything whatever you want. The REF_ in conf daemon knows what the object was so naming an interface Timbuktu is no problem in UTM. Not so in XG and there is cost associated with this because PORT1 is hard coded everywhere, you can't just rename it in the GUI. You can't create Qos rules and then try to rename them. This is basic stuff yet XG cannot do it.

    This is so frustrating...

  • Thanks .

    Whatever the OS is, XG does not match up to our expectations. work? The group where he is working in all the time make a great job (apart in the v17 the content filter windows is too small when you need to edit a web profile) but the XG web section is great and the layout/idea used behind I love it.

    Daemons created in-house? Sophos is trying to look like Fortinet, where they develop everything in house. Fortigate AV does not work and web filtering? Forget it (it misses so many ADS and web content).

    A company can produce a limited number of good products but not all. UTM's strength was the usage of third-party modules developed by others SW houses, like the Web Filtering (McAfee), MTA, etc.

    CLI? I love the CLI and also can give some flexibility and also will cover all the customers that love CLI. This is one of the reason that Customers choose Fortigate. You can create Firewall rules, objects, users, VPN settings from CLI. Their CLI is well-ordered!

    I think that Sophos has the best features sets on the market but they are spending time or bending their head on how to make things work. If you do not have a solid base, your castle will fall down as soon you will add more floors on it (and this is already happening).

    To be honest, I do not have other words to show how frustated and disappointed I am, Bill.

  • When v17b is finished and in GA land I will have to rebuild my test box as a UTM again because so far the lack of fixes is disappointing.

    No IPv6 improvements, country blocking not fixed, MTA still not built correctly, reports well they are still broken, logs might get included in v17 GA or maybe MR1?, clienteles users still require an email address, hang on they are clientless. Waiting until tomorrow to see the change to daylight savings affects the report generation time? No-one appears to be tracking this bug?

    Ian

    XG115W - v20.0.3 MR-3 - Home

    XG on VM 8 - v21 GA

    If a post solves your question please use the 'Verify Answer' button.

  • Ian,

    I can confirm that with v17 I receive daily reports with 2 hours of delay.

  • ChrisKnight said:
    So realistically we're still years away from seeing feature parity with UTM

    My point was not to explain future velocity, and it would be completely misleading if you try to interpret it that way. I only meant to illustrate that new features are typically too expensive to develop during the timeline of a beta.

    We've been working on v17, since v16.5 shipped, with the exclusion of a period of several months, where we focused on increasing our ability release higher quality code. That means that what you see now, was developed during that period, and that's it. That's not to say that all of our engineering capacity was working on v17. Some of our development teams have in that same time, have already been working on the foundations of the next major version of XG, and we'll talk later, about what that brings, but I believe it will answer a number of important criticisms. 

    As for email specifically, you can expect any significant remaining gaps with UTM9, to be closed early 2018.

  • rfcat_vk said:
    country blocking not fixed

    What is broken in country blocking?

  • Speaking more generally to the many other recent replies, I understand that a number of you are disappointed with the improvements in this release. There are two general complaint buckets. One, is that you want more in v17. I was not planning on outlining what's coming next until the end of the beta, but Ill prepare something for later this week, to at least roughly outline what our next step plans are. I think it there is enough concern over feature velocity, that this is needed sooner than I hoped. The other main complaint, is that there are directional concerns, on why we're doing one thing over another. 

    As frustrating as raw feedback can be at times, I do always appreciate it. Rubbing a rough block of wood with silk, won't make the wood any smoother. I'll try to summarize and respond to some of those, in the near future, also. 

     

  • Here the thread:

    https://community.sophos.com/products/xg-firewall/f/firewall-and-policies/76238/country-blocking-not-working-for-a-wan-lan-rule/294125#pi2151=4

    Broken since April 2016 and no one from Sophos gave an official answer on this issue (only few days ago it seems).

    What else we can do AlanT to be heard?

    You should organize a webinar soon and invite some Sophos Partners to explain what you are doing more in depth than a single reply here.

    We are quite disappointed if we look back to 2015 and to this year 2017. In my opinion you should improve the base, first, make it solid and then add layers and layers.

    Regards

  • lferrara said:
    Broken since April 2016 and no one from Sophos gave an official answer on this issue

    yes, this looks like some poor communication, it seems, and I've now discussed this with engineering. It should be done for v17 GA. The problem was that the fix is a rather significant change, and  

    lferrara said:
    In my opinion you should improve the base, first, make it solid and then add layers and layers

    There are two areas that we are currently working on, that are effectively the "base" of the firewall.

    • Configuration and service management
    • UI Framework

    The reason for this, is not that they're unstable or broken, but we believe we can develop features faster, and with higher quality, if we make some significant improvements to these areas. Internally, this is called project Picasso, and development has been underway since before v17 started. Some projects can't be accelerated by doing more in parallel, and they just take time. That doesn't mean we stop innovating until its done, and it doesn't mean what we have today is not ready. We will release v17, then put out a couple more incremental feature updates to v17, before we get to v18, with the some "base" improvements. As I mentioned earlier, I'll outline a bit more about what those are, later this week. 

  • or someone on behalf you should report correctly issue reported here. Another issue is with Bookmarks.

    For the project Picasso, we should see something at the end of next year (I guess) and time goes by and in the meanwhile Sophos is losing credibility.

    XG has some nice and unique Features such as Heartbeat (but here another big issue, it works only with Sophos Cloud) but customers move to soemthing else as soon they discover the basic missing features.

    Make sure you are around more often and not only when a new beta comes out!

    Kind regards,

Reply
  • or someone on behalf you should report correctly issue reported here. Another issue is with Bookmarks.

    For the project Picasso, we should see something at the end of next year (I guess) and time goes by and in the meanwhile Sophos is losing credibility.

    XG has some nice and unique Features such as Heartbeat (but here another big issue, it works only with Sophos Cloud) but customers move to soemthing else as soon they discover the basic missing features.

    Make sure you are around more often and not only when a new beta comes out!

    Kind regards,

Children
  • Hi folks,

    an update to the reporting bug that no-one really seems interested in.

    I updated the bios to the current time as previously reported so the report was being generated at approx 1am AEST. Now, last weekend we moved to AEDT and now the report is generated approx 1 hour later.

     

    Ian

    XG115W - v20.0.3 MR-3 - Home

    XG on VM 8 - v21 GA

    If a post solves your question please use the 'Verify Answer' button.

  • I know this is completely off topic, but I would like to give a shout out to my friend  . He follows XG more closely than most of the sophos staff on this board that like to mark every statement they make as answer to a legitimate question.  , if you want to know what people are generally asking for, ask Luk to send you his recommendations. I bet kudos and praises for XG will increase overnight if you implement even half of the stuff he recommends.

    Good job Luk.

  • Billybob said:

    I know this is completely off topic, but I would like to give a shout out to my friend  . He follows XG more closely than most of the sophos staff on this board that like to mark every statement they make as answer to a legitimate question.  , if you want to know what people are generally asking for, ask Luk to send you his recommendations. I bet kudos and praises for XG will increase overnight if you implement even half of the stuff he recommends.

    Good job Luk.

     

    Thanks Bill. Criticisms are part of our life and our job, in any field. I know I am severe but I always try to be honest with myself and with others things that I want to improve or see improving. I am really frustating to see how the Astaro project has been destroyed (look at the UTM bugs in the last year).

    I am here almost everyday as soon I have 5 minutes I am on the community to help other people and to give my feeedback to Sophos but in the last few months I realized that Copernicus project is not the one I would expected at the beginning. I really hope that in Sophos there will be a change of course. AlanT did a great job of managing UTM and improving astaro from version 8 to 9.

    v18 and 2018 is the final call for XG in my opinion. Sophos should give gifts to current XG customers so that they will not lose them. We expect to see a feedback form here on community or by Partner portal as they did after the v15.

    Mine are ideas but maybe they have better ideas than mine.

  • Same thing for me Luk... If v18 is OK well and fix the gap well, otherwise I'm already ready for an alternative and migrate all my clients to another vendor ... and I'm sure I will not be the only one to think that way!

    AlanT ask to Luk!!