写在前面:
在学习蜜罐的相关内容时,接触了一些英文资料,对这些英文资料的翻译在个人博客中贴出,希望我的有限英文水平能够为大家做一点点贡献。如有翻译不妥的地方,大家可以留言指正,共同进步。
做快乐的程序员,做热爱分享的程序员。
请大家不要纠结某个词、字的使用合不合理,理解论文主要讲述的内容为先。
每篇文章版权都会在原始文章最后注明,请大家遵守。
标签:蜜罐,云计算,IaaS
For the evaluation ofthe proposed architecture, we use a modified Xen 4.1.2 version and a hardwarenode with an Intel(R) Xeon(R) Quad CPU X3450 with 2.67 GHz and 4 GB of ram.
The Xen dom0 runs Ubuntu Linux 10.04 (i386)and kernel 3.0.013-generic-pae.
The domU guest VM runs Ubuntu Linux 11.10 (i386) with kernel 3.0.019genericpae.
运行的一些版本信息
Since the snapshot only saves changes on the file p_w_picpath,thecreation process of a new lvm2 snapshot in our setup is very fast .
由于快照仅保存在文件映射上的更改,我们设置一个新的 lvm2 快照的创建过程是非常快的。
First, the passwd and shadow files are replaced withmodified files that disable the opportunity to log in.
首先,通过修改文件将 passwd 和shadow文件替换,目的是禁用登录。
Second, the web server data is removed and defined sectorsare blocked.
第二,删除了 web 服务器数据并阻止定义的扇区的访问。
The sector blocking has minimal performance overhead.
封闭扇形具有最小的性能开销。
This was optimized using requests to a sorted list offorbidden sectors and a binary search algorithm.
最佳的选择是使用,禁止的扇区的排序列表和二叉树搜索算法。
The time needed for the extraction is mostly influenced bythe live cloning process and the copy of the memory of the original VM (Table4).
提取所需的时间,主要受活克隆过程和原始 VM的内存的拷贝影响。
Since we deploy a honeypot only if needed and do not occupyprepared resources, these timings are good results.
既然我们仅在需要时才部署一个蜜罐,并不会占用准备的资源,这些计时都是很好的结果。
These timings are measured during the original VM is at lowcapacity utilization.
这些计时是在原始 VM 处于低容量利用率时测量的。
Due to the live migration algorithm, the more memory pagesare modified during the copy procedure, the more copy iterations have to beperformed.If the VM is highly utilized, the live cloning procedure takes muchmore time .
由于实时迁移算法,在复制过程中更多的内存页被修改,更多的副本迭代必须执行。如果 VM 的利用率高,活的克隆过程需要更多的时间。
In order to improve this, we modified the live migrationalgorithm of the Xen hypervisor to minimize performed copy iterations andmaximized amount of dirty memory pages which are directly copied.
为了改善这种情况,我们修改 Xen 虚拟机监控程序的动态迁移算法,以尽量减少执行的复制的迭代,并最大化脏内存页面直接复制的数量。
This results in a constantly fast live cloning procedureeven if the VM is highly utilized.
这导致了,倘若 VM 利用率高,会不断进行快速活的克隆。
Disadvantage is that the honeypot VM is not fullysynchronized with the original VM, but this is acceptable for ourpurposes.Thereby, the extraction procedure stays constant in time.
缺点是蜜罐 VM 与原始的 VM,不完全同步,但这对于我们的目的是可以接受的。从而,萃取过程保持在不变的时间内。
The honeypot VM must be closely monitored in order torespond quickly to successful attacks.
蜜罐 VM 必须密切监测,以快速响应成功的***。
If a successful attack on the honeypot has been detected, itmust be immediately terminated and no outgoing traffic should be able to leavein any way.
如果已检测到对蜜罐技术成功的***,必须立即终止,并没有传出通信应该能够以任何方式离开。
Retrieving information using VMI and comparing 80 runningprocesses to the previous information retrieval step takes about 66±11 ms.
使用VMI和对以前的信息检索步骤中80个正在运行的进程进行比较,检索信息需要大约 66±11毫秒。
Retrieving information about 14 loaded kernel modules can beperformed in 54 ±7 ms in our setup.
在我们的设置中,可以在54 ± 7 ms 内执行 14条加载的内核模块的检索信息
Additionally, the controller can monitor files by examiningthe raw p_w_picpath file of the honeypot VM.
此外,该控制器可以通过检查蜜罐 VM 的原始映像文件,监视文件。
It can determine, if certain files are accessed or modified.
它可以确定是否某些文件被访问或修改。
Table 5 shows the timings for retrieving information aboutsome files on a honeypot VM.
表 5 显示在一些蜜罐 VM内某些文件的检索信息的检索时间。
All in all, the honeypot controller can continuously monitorand compare every 250 ms the processes and modules of the running honeypot VMand check 12 files on the raw p_w_picpath snapshot for modifications.
总的来说,蜜罐控制器可以连续监控,比较每个 250 毫秒内运行蜜罐 VM的进程和模块,检查 12个原始映像快照文件中的修改。
为方便大家,我会把总的doc文档上传到网盘
百度文库
51cto:http://down.51cto.com/data/1887779
csdn
百度网盘:
个人博客:http://fergusj.blog.51cto.com
如果您喜欢这个文章,读后觉得收获很大,不妨留言点赞,让我有动力继续写出高质量的翻译。
免责声明:本站发布的内容(图片、视频和文字)以原创、转载和分享为主,文章观点不代表本网站立场,如果涉及侵权请联系站长邮箱:is@yisu.com进行举报,并提供相关证据,一经查实,将立刻删除涉嫌侵权内容。